[{"_1":2,"_457":-5,"_458":-5},"loaderData",{"_3":4,"_24":25,"_345":346},"root",{"_5":6,"_7":8},"toasts",[],"visitor",{"_9":10,"_11":12},"viewportWidth",1280,"visitorSession",{"_13":14,"_16":17,"_18":-5,"_19":20,"_21":22,"_23":-5},"attributes",[15],"bot","id",27182804,"membership","referenceId","019ab33c-32bd-768e-80e6-9ee61bfabc38","unreadReleaseNotesCount",2,"userAccount","routes/_public/mcp/servers/~namespace/~slug/_pages/_layout",{"_26":27,"_83":-5,"_84":41,"_29":85,"_262":263,"_343":344},"availableLocales",[28,41,48,55,62,69,76],{"_29":30,"_37":38,"_39":40},"mcpServer",{"_31":32,"_33":34,"_35":36},"description","Ermöglicht KI-Agenten die Interaktion mit Webbrowsern unter Verwendung natürlicher Sprache und bietet automatisiertes Browsen, Ausfüllen von Formularen, visuelle Elementerkennung und strukturierte JSON-Antworten für eine systematische Browsersteuerung.","displayName","MCP-Browser-Nutzungsserver","readme","# MCP-Server mit Browsernutzung\n\n[![Schmiedeabzeichen](https://smithery.ai/badge/@JovaniPink/mcp-browser-use)](https://smithery.ai/server/@JovaniPink/mcp-browser-use)\n\n> MCP-Server zur [Browsernutzung](https://github.com/browser-use/browser-use) .\n\n## Überblick\n\nDieses Repository enthält den Server für die [Browser-Use](https://github.com/browser-use/browser-use) -Bibliothek. Diese stellt ein leistungsstarkes Browser-Automatisierungssystem bereit, das KI-Agenten die Interaktion mit Webbrowsern über natürliche Sprache ermöglicht. Der Server basiert auf [dem Model Context Protocol (MCP)](https://modelcontextprotocol.io/introduction) von Anthropic und bietet eine nahtlose Integration mit der [Browser-Use-](https://github.com/browser-use/browser-use) Bibliothek.\n\n## Merkmale\n\n1. **Browsersteuerung**\n\n* Automatisierte Browserinteraktionen über natürliche Sprache\n* Navigations-, Formularausfüll-, Klick- und Scrollfunktionen\n* Tab-Verwaltung und Screenshot-Funktionalität\n* Cookie- und Statusverwaltung\n\n2. **Agentensystem**\n\n* Benutzerdefinierte Agentimplementierung in custom\\_agent.py\n* Bildbasierte Elementerkennung\n* Strukturierte JSON-Antworten für Aktionen\n* Verwaltung und Zusammenfassung des Nachrichtenverlaufs\n\n3. **Konfiguration**\n\n* Umgebungsbasierte Konfiguration für API-Schlüssel und -Einstellungen\n* Chrome-Browsereinstellungen (Debug-Port, Persistenz)\n* Modellanbieterauswahl und Parameter\n\n## Abhängigkeiten\n\nDieses Projekt basiert auf den folgenden Python-Paketen:\n\n| Paket | Version | Beschreibung |\n| :---------------------------------------------------------------------------------------------------------- | :-------- | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| [Kissen](https://python-pillow.org/) | >=10.1.0 | Python Imaging Library (PIL)-Fork, der Ihrem Python-Interpreter Bildverarbeitungsfunktionen hinzufügt. |\n| [Browser-Nutzung](https://github.com/browser-use/browser-use) | ==0.1.19 | Ein leistungsstarkes Browser-Automatisierungssystem, das KI-Agenten die Interaktion mit Webbrowsern über natürliche Sprache ermöglicht. Die Kernbibliothek, die die Browser-Automatisierungsfunktionen dieses Projekts unterstützt. |\n| [Fastapi](https://fastapi.tiangolo.com/) | >=0,115,6 | Modernes, schnelles (leistungsstarkes) Webframework zum Erstellen von APIs mit Python 3.7+ basierend auf standardmäßigen Python-Typhinweisen. Wird zum Erstellen des Servers verwendet, der die Funktionalität des Agenten bereitstellt. |\n| [fastmcp](https://pypi.org/project/fastmcp/) | >=0,4,1 | Ein Framework, das FastAPI zum Erstellen von MCP-Servern (Model Context Protocol) umschließt. |\n| [Lehrer](https://github.com/jxnl/instructor) | >=1,7,2 | Bibliothek zur strukturierten Ausgabeaufforderung und Validierung mit OpenAI-Modellen. Ermöglicht das Extrahieren strukturierter Daten aus Modellantworten. |\n| [Langkette](https://www.langchain.com/) | >=0,3,14 | Framework für die Entwicklung von Anwendungen mit großen Sprachmodellen (LLMs). Bietet Tools zum Verketten verschiedener Sprachmodellkomponenten und zur Interaktion mit verschiedenen APIs und Datenquellen. |\n| [langchain-google-genai](https://pypi.org/project/langchain-google-genai/) | >=2.1.1 | LangChain-Integration für Google GenAI-Modelle, die die Nutzung der generativen KI-Funktionen von Google innerhalb des LangChain-Frameworks ermöglicht. |\n| [Langchain-Openai](https://api.python.langchain.com/en/latest/langchain_openai.html) | >=0,2,14 | LangChain-Integrationen mit OpenAI-Modellen. Ermöglicht die Verwendung von OpenAI-Modellen (wie GPT-4) innerhalb des LangChain-Frameworks. Wird in diesem Projekt für die Interaktion mit den Sprach- und Bildmodellen von OpenAI verwendet. |\n| [Langkette-Ollama](https://api.python.langchain.com/en/latest/langchain_ollama/chat_models/ChatOllama.html) | >=0,2,2 | Langchain-Integration für Ollama, die die lokale Ausführung von LLMs ermöglicht. |\n| [openai](https://platform.openai.com/docs/api-reference) | >=1,59,5 | Offizielle Python-Clientbibliothek für die OpenAI-API. Wird zur direkten Interaktion mit OpenAI-Modellen verwendet (bei Bedarf zusätzlich zu LangChain). |\n| [python-dotenv](https://pypi.org/project/python-dotenv/) | >=1.0.1 | Liest Schlüssel-Wert-Paare aus einer `.env` Datei und setzt sie als Umgebungsvariablen. Vereinfacht die lokale Entwicklung und das Konfigurationsmanagement. |\n| [pydantisch](https://docs.pydantic.dev/) | >=2.10.5 | Datenvalidierung und Einstellungsverwaltung mithilfe von Python-Typannotationen. Ermöglicht die Laufzeitdurchsetzung von Typen und die automatische Modellerstellung. Unverzichtbar für die Definition strukturierter Datenmodelle im Agenten. |\n| [pyperclip](https://pyperclip.readthedocs.io/) | >=1.9.0 | Plattformübergreifendes Python-Modul zum Kopieren und Einfügen der Zwischenablagefunktionen. |\n| [Uvicorn](https://www.uvicorn.org/) | >=0,22,0 | ASGI-Webserverimplementierung für Python. Wird zur Bereitstellung der FastAPI-Anwendung verwendet. |\n\n## Komponenten\n\n### Ressourcen\n\nDer Server implementiert ein Browser-Automatisierungssystem mit:\n\n* Integration mit Browser-Nutzungsbibliothek für erweiterte Browsersteuerung\n* Benutzerdefinierte Browser-Automatisierungsfunktionen\n* Agentenbasiertes Interaktionssystem mit Vision-Funktionen\n* Persistente Zustandsverwaltung\n* Anpassbare Modelleinstellungen\n\n### Anforderungen\n\n* Betriebssysteme (Linux, macOS, Windows; wir haben nicht für Docker oder Microsoft WSL getestet)\n* Python 3.11 oder höher\n* uv (schneller Python-Paketinstaller)\n* Chrome/Chromium-Browser\n* [Claude Desktop](https://claude.ai/download)\n\n### Schnellstart\n\n#### Claude Desktop\n\nUnter MacOS: `~/Library/Application\\ Support/Claude/claude_desktop_config.json` Unter Windows: `%APPDATA%/Claude/claude_desktop_config.json`\n\n#### Installation über Smithery\n\nSo installieren Sie Browser Use für Claude Desktop automatisch über [Smithery](https://smithery.ai/server/@JovaniPink/mcp-browser-use) :\n\n```bash\nnpx -y @smithery/cli install @JovaniPink/mcp-browser-use --client claude\n```\n\n```json\n\"mcpServers\": {\n \"mcp_server_browser_use\": {\n \"command\": \"uvx\",\n \"args\": [\n \"mcp-server-browser-use\",\n ],\n \"env\": {\n \"OPENAI_ENDPOINT\": \"https://api.openai.com/v1\",\n \"OPENAI_API_KEY\": \"\",\n \"ANTHROPIC_API_KEY\": \"\",\n \"GOOGLE_API_KEY\": \"\",\n \"AZURE_OPENAI_ENDPOINT\": \"\",\n \"AZURE_OPENAI_API_KEY\": \"\",\n // \"DEEPSEEK_ENDPOINT\": \"https://api.deepseek.com\",\n // \"DEEPSEEK_API_KEY\": \"\",\n // Set to false to disable anonymized telemetry\n \"ANONYMIZED_TELEMETRY\": \"false\",\n // Chrome settings\n \"CHROME_PATH\": \"\",\n \"CHROME_USER_DATA\": \"\",\n \"CHROME_DEBUGGING_PORT\": \"9222\",\n \"CHROME_DEBUGGING_HOST\": \"localhost\",\n // Set to true to keep browser open between AI tasks\n \"CHROME_PERSISTENT_SESSION\": \"false\",\n // Model settings\n \"MCP_MODEL_PROVIDER\": \"anthropic\",\n \"MCP_MODEL_NAME\": \"claude-3-5-sonnet-20241022\",\n \"MCP_TEMPERATURE\": \"0.3\",\n \"MCP_MAX_STEPS\": \"30\",\n \"MCP_USE_VISION\": \"true\",\n \"MCP_MAX_ACTIONS_PER_STEP\": \"5\",\n \"MCP_TOOL_CALL_IN_CONTENT\": \"true\"\n }\n }\n}\n```\n\n### Umgebungsvariablen\n\nWichtige Umgebungsvariablen:\n\n```bash\n# API Keys\nANTHROPIC_API_KEY=anthropic_key\n\n# Chrome Configuration\n# Optional: Path to Chrome executable\nCHROME_PATH=/path/to/chrome\n# Optional: Chrome user data directory\nCHROME_USER_DATA=/path/to/user/data\n# Default: 9222\nCHROME_DEBUGGING_PORT=9222\n# Default: localhost\nCHROME_DEBUGGING_HOST=localhost\n# Keep browser open between tasks\nCHROME_PERSISTENT_SESSION=false\n\n# Model Settings\n# Options: anthropic, openai, azure, deepseek\nMCP_MODEL_PROVIDER=anthropic\n# Model name\nMCP_MODEL_NAME=claude-3-5-sonnet-20241022\nMCP_TEMPERATURE=0.3\nMCP_MAX_STEPS=30\nMCP_USE_VISION=true\nMCP_MAX_ACTIONS_PER_STEP=5\n```\n\n## Entwicklung\n\n### Aufstellen\n\n1. Klonen Sie das Repository:\n\n```bash\ngit clone https://github.com/JovaniPink/mcp-browser-use.git\ncd mcp-browser-use\n```\n\n2. Virtuelle Umgebung erstellen und aktivieren:\n\n```bash\npython -m venv .venv\nsource .venv/bin/activate # On Windows: .venv\\Scripts\\activate\n```\n\n3. Installieren Sie Abhängigkeiten:\n\n```bash\nuv sync\n```\n\n4. Starten Sie den Server\n\n```bash\nuv run mcp-browser-use\n```\n\n### Debuggen\n\nVerwenden Sie zum Debuggen den [MCP Inspector](https://github.com/modelcontextprotocol/inspector) :\n\n```bash\nnpx @modelcontextprotocol/inspector uv --directory /path/to/project run mcp-server-browser-use\n```\n\nDer Inspector zeigt eine URL für die Debugging-Schnittstelle an.\n\n## Browseraktionen\n\nDer Server unterstützt verschiedene Browseraktionen durch natürliche Sprache:\n\n* Navigation: Zu URLs gehen, zurück/vorwärts, aktualisieren\n* Interaktion: Klicken, tippen, scrollen, schweben\n* Formulare: Formulare ausfüllen, absenden, Optionen auswählen\n* Status: Seiteninhalt abrufen, Screenshots machen\n* Tabs: Tabs erstellen, schließen, zwischen Tabs wechseln\n* Vision: Elemente anhand der visuellen Erscheinung finden\n* Cookies und Speicher: Browserstatus verwalten\n\n## Sicherheit\n\nIch möchte darauf hinweisen, dass einige Chrome-Einstellungen die Steuerung des Browsers durch den Server ermöglichen. Dies stellt ein Sicherheitsrisiko dar und sollte mit Vorsicht verwendet werden. Der Server ist nicht für den Einsatz in einer Produktionsumgebung vorgesehen.\n\nSicherheitsdetails: [SECURITY.MD](./documentation/SECURITY.md)\n\n## Beitragen\n\nWir freuen uns über Beiträge zu diesem Projekt. Bitte folgen Sie diesen Schritten:\n\n1. Forken Sie dieses Repository.\n2. Erstellen Sie Ihren Feature-Zweig: `git checkout -b my-new-feature` .\n3. Übernehmen Sie Ihre Änderungen: `git commit -m 'Add some feature'` .\n4. Pushen zum Zweig: `git push origin my-new-feature` .\n5. Senden Sie eine Pull-Anfrage.\n\nBei größeren Änderungen öffnen Sie zunächst ein Problem, um zu besprechen, was Sie ändern möchten. Bitte aktualisieren Sie die Tests entsprechend, um die vorgenommenen Änderungen zu berücksichtigen.","name","German","tag","de-DE",{"_29":42,"_37":46,"_39":47},{"_31":43,"_33":44,"_35":45},"Enables AI agents to interact with web browsers using natural language, featuring automated browsing, form filling, vision-based element detection, and structured JSON responses for systematic browser control.","MCP Browser Use Server","# MCP server w/ Browser Use\n\n[![smithery badge](https://smithery.ai/badge/@JovaniPink/mcp-browser-use)](https://smithery.ai/server/@JovaniPink/mcp-browser-use)\n\n> MCP server for [browser-use](https://github.com/browser-use/browser-use).\n\n $\"Browser$ \n\n## Overview\n\nThis repository contains the server for the [browser-use](https://github.com/browser-use/browser-use) library, which provides a powerful browser automation system that enables AI agents to interact with web browsers through natural language. The server is built on Anthropic's [Model Context Protocol (MCP)](https://modelcontextprotocol.io/introduction) and provides a seamless integration with the [browser-use](https://github.com/browser-use/browser-use) library.\n\n## Features\n\n1. **Browser Control**\n\n- Automated browser interactions via natural language\n- Navigation, form filling, clicking, and scrolling capabilities\n- Tab management and screenshot functionality\n- Cookie and state management\n\n2. **Agent System**\n\n- Custom agent implementation in custom_agent.py\n- Vision-based element detection\n- Structured JSON responses for actions\n- Message history management and summarization\n\n3. **Configuration**\n\n- Environment-based configuration for API keys and settings\n- Chrome browser settings (debugging port, persistence)\n- Model provider selection and parameters\n\n## Dependencies\n\nThis project relies on the following Python packages:\n\n| Package | Version | Description |\n| :----------------------------------------- | :--------- | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |\n| [Pillow](https://python-pillow.org/) | >=10.1.0 | Python Imaging Library (PIL) fork that adds image processing capabilities to your Python interpreter. |\n| [browser-use](https://github.com/browser-use/browser-use) | ==0.1.19 | A powerful browser automation system that enables AI agents to interact with web browsers through natural language. The core library that powers this project's browser automation capabilities. |\n| [fastapi](https://fastapi.tiangolo.com/) | >=0.115.6 | Modern, fast (high-performance) web framework for building APIs with Python 3.7+ based on standard Python type hints. Used to create the server that exposes the agent's functionality. |\n| [fastmcp](https://pypi.org/project/fastmcp/) | >=0.4.1 | A framework that wraps FastAPI for building MCP (Model Context Protocol) servers. |\n| [instructor](https://github.com/jxnl/instructor) | >=1.7.2 | Library for structured output prompting and validation with OpenAI models. Enables extracting structured data from model responses. |\n| [langchain](https://www.langchain.com/) | >=0.3.14 | Framework for developing applications with large language models (LLMs). Provides tools for chaining together different language model components and interacting with various APIs and data sources. |\n| [langchain-google-genai](https://pypi.org/project/langchain-google-genai/) | >=2.1.1 | LangChain integration for Google GenAI models, enabling the use of Google's generative AI capabilities within the LangChain framework. |\n| [langchain-openai](https://api.python.langchain.com/en/latest/langchain_openai.html) | >=0.2.14 | LangChain integrations with OpenAI's models. Enables using OpenAI models (like GPT-4) within the LangChain framework. Used in this project for interacting with OpenAI's language and vision models. |\n| [langchain-ollama](https://api.python.langchain.com/en/latest/langchain_ollama/chat_models/ChatOllama.html) | >=0.2.2 | Langchain integration for Ollama, enabling local execution of LLMs. |\n| [openai](https://platform.openai.com/docs/api-reference) | >=1.59.5 | Official Python client library for the OpenAI API. Used to interact directly with OpenAI's models (if needed, in addition to LangChain). |\n| [python-dotenv](https://pypi.org/project/python-dotenv/) | >=1.0.1 | Reads key-value pairs from a `.env` file and sets them as environment variables. Simplifies local development and configuration management. |\n| [pydantic](https://docs.pydantic.dev/) | >=2.10.5 | Data validation and settings management using Python type annotations. Provides runtime enforcement of types and automatic model creation. Essential for defining structured data models in the agent. |\n| [pyperclip](https://pyperclip.readthedocs.io/) | >=1.9.0 | Cross-platform Python module for copy and paste clipboard functions. |\n| [uvicorn](https://www.uvicorn.org/) | >=0.22.0 | ASGI web server implementation for Python. Used to serve the FastAPI application. |\n\n## Components\n\n### Resources\n\nThe server implements a browser automation system with:\n\n- Integration with browser-use library for advanced browser control\n- Custom browser automation capabilities\n- Agent-based interaction system with vision capabilities\n- Persistent state management\n- Customizable model settings\n\n### Requirements\n\n- Operating Systems (Linux, macOS, Windows; we haven't tested for Docker or Microsoft WSL)\n- Python 3.11 or higher\n- uv (fast Python package installer)\n- Chrome/Chromium browser\n- [Claude Desktop](https://claude.ai/download)\n\n### Quick Start\n\n#### Claude Desktop\n\nOn MacOS: `~/Library/Application\\ Support/Claude/claude_desktop_config.json`\nOn Windows: `%APPDATA%/Claude/claude_desktop_config.json`\n\n#### Installing via Smithery\n\nTo install Browser Use for Claude Desktop automatically via [Smithery](https://smithery.ai/server/@JovaniPink/mcp-browser-use):\n\n```bash\nnpx -y @smithery/cli install @JovaniPink/mcp-browser-use --client claude\n```\n\n

Development Configuration

\n\n```json\n\"mcpServers\": {\n \"mcp_server_browser_use\": {\n \"command\": \"uvx\",\n \"args\": [\n \"mcp-server-browser-use\",\n ],\n \"env\": {\n \"OPENAI_ENDPOINT\": \"https://api.openai.com/v1\",\n \"OPENAI_API_KEY\": \"\",\n \"ANTHROPIC_API_KEY\": \"\",\n \"GOOGLE_API_KEY\": \"\",\n \"AZURE_OPENAI_ENDPOINT\": \"\",\n \"AZURE_OPENAI_API_KEY\": \"\",\n // \"DEEPSEEK_ENDPOINT\": \"https://api.deepseek.com\",\n // \"DEEPSEEK_API_KEY\": \"\",\n // Set to false to disable anonymized telemetry\n \"ANONYMIZED_TELEMETRY\": \"false\",\n // Chrome settings\n \"CHROME_PATH\": \"\",\n \"CHROME_USER_DATA\": \"\",\n \"CHROME_DEBUGGING_PORT\": \"9222\",\n \"CHROME_DEBUGGING_HOST\": \"localhost\",\n // Set to true to keep browser open between AI tasks\n \"CHROME_PERSISTENT_SESSION\": \"false\",\n // Model settings\n \"MCP_MODEL_PROVIDER\": \"anthropic\",\n \"MCP_MODEL_NAME\": \"claude-3-5-sonnet-20241022\",\n \"MCP_TEMPERATURE\": \"0.3\",\n \"MCP_MAX_STEPS\": \"30\",\n \"MCP_USE_VISION\": \"true\",\n \"MCP_MAX_ACTIONS_PER_STEP\": \"5\",\n \"MCP_TOOL_CALL_IN_CONTENT\": \"true\"\n }\n }\n}\n```\n\n

\n\n### Environment Variables\n\nKey environment variables:\n\n```bash\n# API Keys\nANTHROPIC_API_KEY=anthropic_key\n\n# Chrome Configuration\n# Optional: Path to Chrome executable\nCHROME_PATH=/path/to/chrome\n# Optional: Chrome user data directory\nCHROME_USER_DATA=/path/to/user/data\n# Default: 9222\nCHROME_DEBUGGING_PORT=9222\n# Default: localhost\nCHROME_DEBUGGING_HOST=localhost\n# Keep browser open between tasks\nCHROME_PERSISTENT_SESSION=false\n\n# Model Settings\n# Options: anthropic, openai, azure, deepseek\nMCP_MODEL_PROVIDER=anthropic\n# Model name\nMCP_MODEL_NAME=claude-3-5-sonnet-20241022\nMCP_TEMPERATURE=0.3\nMCP_MAX_STEPS=30\nMCP_USE_VISION=true\nMCP_MAX_ACTIONS_PER_STEP=5\n```\n\n## Development\n\n### Setup\n\n1. Clone the repository:\n\n```bash\ngit clone https://github.com/JovaniPink/mcp-browser-use.git\ncd mcp-browser-use\n```\n\n2. Create and activate virtual environment:\n\n```bash\npython -m venv .venv\nsource .venv/bin/activate # On Windows: .venv\\Scripts\\activate\n```\n\n3. Install dependencies:\n\n```bash\nuv sync\n```\n\n4. Start the server\n\n```bash\nuv run mcp-browser-use\n```\n\n### Debugging\n\nFor debugging, use the [MCP Inspector](https://github.com/modelcontextprotocol/inspector):\n\n```bash\nnpx @modelcontextprotocol/inspector uv --directory /path/to/project run mcp-server-browser-use\n```\n\nThe Inspector will display a URL for the debugging interface.\n\n## Browser Actions\n\nThe server supports various browser actions through natural language:\n\n- Navigation: Go to URLs, back/forward, refresh\n- Interaction: Click, type, scroll, hover\n- Forms: Fill forms, submit, select options\n- State: Get page content, take screenshots\n- Tabs: Create, close, switch between tabs\n- Vision: Find elements by visual appearance\n- Cookies & Storage: Manage browser state\n\n## Security\n\nI want to note that their are some Chrome settings that are set to allow for the browser to be controlled by the server. This is a security risk and should be used with caution. The server is not intended to be used in a production environment.\n\nSecurity Details: [SECURITY.MD](./documentation/SECURITY.md)\n\n## Contributing\n\nWe welcome contributions to this project. Please follow these steps:\n\n1. Fork this repository.\n2. Create your feature branch: `git checkout -b my-new-feature`.\n3. Commit your changes: `git commit -m 'Add some feature'`.\n4. Push to the branch: `git push origin my-new-feature`.\n5. Submit a pull request.\n\nFor major changes, open an issue first to discuss what you would like to change. Please update tests as appropriate to reflect any changes made.\n","English","en-US",{"_29":49,"_37":53,"_39":54},{"_31":50,"_33":51,"_35":52},"Permite que los agentes de IA interactúen con los navegadores web utilizando lenguaje natural, presentando navegación automatizada, llenado de formularios, detección de elementos basada en visión y respuestas JSON estructuradas para el control sistemático del navegador.","Servidor de uso del navegador MCP","# Servidor MCP con uso del navegador\n\n[![insignia de herrería](https://smithery.ai/badge/@JovaniPink/mcp-browser-use)](https://smithery.ai/server/@JovaniPink/mcp-browser-use)\n\n> Servidor MCP para [uso del navegador](https://github.com/browser-use/browser-use) .\n\n## Descripción general\n\nEste repositorio contiene el servidor para la biblioteca [de uso del navegador](https://github.com/browser-use/browser-use) , que proporciona un potente sistema de automatización del navegador que permite a los agentes de IA interactuar con los navegadores web mediante lenguaje natural. El servidor se basa en [el Protocolo de Contexto de Modelo (MCP)](https://modelcontextprotocol.io/introduction) de Anthropic y se integra perfectamente con la biblioteca [de uso del navegador](https://github.com/browser-use/browser-use) .\n\n## Características\n\n1. **Control del navegador**\n\n* Interacciones automatizadas del navegador a través del lenguaje natural\n* Capacidades de navegación, llenado de formularios, clics y desplazamiento\n* Gestión de pestañas y funcionalidad de captura de pantalla\n* Gestión de cookies y estados\n\n2. **Sistema de agentes**\n\n* Implementación de agente personalizado en custom\\_agent.py\n* Detección de elementos basada en visión\n* Respuestas JSON estructuradas para acciones\n* Gestión y resumen del historial de mensajes\n\n3. **Configuración**\n\n* Configuración basada en el entorno para claves y configuraciones de API\n* Configuración del navegador Chrome (puerto de depuración, persistencia)\n* Selección del proveedor del modelo y parámetros\n\n## Dependencias\n\nEste proyecto se basa en los siguientes paquetes de Python:\n\n| Paquete | Versión | Descripción |\n| :---------------------------------------------------------------------------------------------------------- | :-------- | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| [Almohada](https://python-pillow.org/) | >=10.1.0 | Bifurcación de Python Imaging Library (PIL) que agrega capacidades de procesamiento de imágenes a su intérprete de Python. |\n| [uso del navegador](https://github.com/browser-use/browser-use) | ==0.1.19 | Un potente sistema de automatización de navegadores que permite a los agentes de IA interactuar con los navegadores web mediante lenguaje natural. La biblioteca principal que impulsa las capacidades de automatización de navegadores de este proyecto. |\n| [Fastapi](https://fastapi.tiangolo.com/) | >=0.115.6 | Framework web moderno, rápido y de alto rendimiento para crear API con Python 3.7+, basado en sugerencias de tipos estándar de Python. Se utiliza para crear el servidor que expone la funcionalidad del agente. |\n| [fastmcp](https://pypi.org/project/fastmcp/) | >=0,4,1 | Un marco que envuelve FastAPI para crear servidores MCP (Protocolo de contexto de modelo). |\n| [instructor](https://github.com/jxnl/instructor) | >=1.7.2 | Biblioteca para la generación y validación de resultados estructurados con modelos OpenAI. Permite extraer datos estructurados de las respuestas de los modelos. |\n| [cadena larga](https://www.langchain.com/) | >=0,3,14 | Marco para el desarrollo de aplicaciones con grandes modelos de lenguaje (LLM). Proporciona herramientas para encadenar diferentes componentes de modelos de lenguaje e interactuar con diversas API y fuentes de datos. |\n| [langchain-google-genai](https://pypi.org/project/langchain-google-genai/) | >=2.1.1 | Integración de LangChain con los modelos GenAI de Google, que permite el uso de las capacidades de inteligencia artificial generativa de Google dentro del marco de LangChain. |\n| [langchain-openai](https://api.python.langchain.com/en/latest/langchain_openai.html) | >=0,2,14 | Integración de LangChain con los modelos de OpenAI. Permite usar modelos de OpenAI (como GPT-4) dentro del framework LangChain. Se utiliza en este proyecto para interactuar con los modelos de lenguaje y visión de OpenAI. |\n| [langchain-ollama](https://api.python.langchain.com/en/latest/langchain_ollama/chat_models/ChatOllama.html) | >=0,2,2 | Integración de Langchain para Ollama, que permite la ejecución local de LLM. |\n| [openai](https://platform.openai.com/docs/api-reference) | >=1,59,5 | Biblioteca cliente oficial de Python para la API de OpenAI. Se utiliza para interactuar directamente con los modelos de OpenAI (si es necesario, además de LangChain). |\n| [python-dotenv](https://pypi.org/project/python-dotenv/) | >=1.0.1 | Lee pares clave-valor de un archivo `.env` y los configura como variables de entorno. Simplifica el desarrollo local y la gestión de la configuración. |\n| [pidantico](https://docs.pydantic.dev/) | >=2.10.5 | Validación de datos y gestión de configuraciones mediante anotaciones de tipos de Python. Permite la aplicación de tipos en tiempo de ejecución y la creación automática de modelos. Esencial para definir modelos de datos estructurados en el agente. |\n| [clip de pyper](https://pyperclip.readthedocs.io/) | >=1.9.0 | Módulo Python multiplataforma para funciones de copiar y pegar el portapapeles. |\n| [uvicornio](https://www.uvicorn.org/) | >=0,22,0 | Implementación del servidor web ASGI para Python. Se utiliza para la aplicación FastAPI. |\n\n## Componentes\n\n### Recursos\n\nEl servidor implementa un sistema de automatización del navegador con:\n\n* Integración con la biblioteca de uso del navegador para un control avanzado del navegador\n* Capacidades de automatización del navegador personalizado\n* Sistema de interacción basado en agentes con capacidades de visión\n* Gestión de estados persistentes\n* Configuraciones de modelo personalizables\n\n### Requisitos\n\n* Sistemas operativos (Linux, macOS, Windows; no hemos realizado pruebas con Docker ni Microsoft WSL)\n* Python 3.11 o superior\n* uv (instalador rápido de paquetes de Python)\n* Navegador Chrome/Chromium\n* [Escritorio de Claude](https://claude.ai/download)\n\n### Inicio rápido\n\n#### Escritorio de Claude\n\nEn MacOS: `~/Library/Application\\ Support/Claude/claude_desktop_config.json` En Windows: `%APPDATA%/Claude/claude_desktop_config.json`\n\n#### Instalación mediante herrería\n\nPara instalar Browser Use for Claude Desktop automáticamente a través de [Smithery](https://smithery.ai/server/@JovaniPink/mcp-browser-use) :\n\n```bash\nnpx -y @smithery/cli install @JovaniPink/mcp-browser-use --client claude\n```\n\n```json\n\"mcpServers\": {\n \"mcp_server_browser_use\": {\n \"command\": \"uvx\",\n \"args\": [\n \"mcp-server-browser-use\",\n ],\n \"env\": {\n \"OPENAI_ENDPOINT\": \"https://api.openai.com/v1\",\n \"OPENAI_API_KEY\": \"\",\n \"ANTHROPIC_API_KEY\": \"\",\n \"GOOGLE_API_KEY\": \"\",\n \"AZURE_OPENAI_ENDPOINT\": \"\",\n \"AZURE_OPENAI_API_KEY\": \"\",\n // \"DEEPSEEK_ENDPOINT\": \"https://api.deepseek.com\",\n // \"DEEPSEEK_API_KEY\": \"\",\n // Set to false to disable anonymized telemetry\n \"ANONYMIZED_TELEMETRY\": \"false\",\n // Chrome settings\n \"CHROME_PATH\": \"\",\n \"CHROME_USER_DATA\": \"\",\n \"CHROME_DEBUGGING_PORT\": \"9222\",\n \"CHROME_DEBUGGING_HOST\": \"localhost\",\n // Set to true to keep browser open between AI tasks\n \"CHROME_PERSISTENT_SESSION\": \"false\",\n // Model settings\n \"MCP_MODEL_PROVIDER\": \"anthropic\",\n \"MCP_MODEL_NAME\": \"claude-3-5-sonnet-20241022\",\n \"MCP_TEMPERATURE\": \"0.3\",\n \"MCP_MAX_STEPS\": \"30\",\n \"MCP_USE_VISION\": \"true\",\n \"MCP_MAX_ACTIONS_PER_STEP\": \"5\",\n \"MCP_TOOL_CALL_IN_CONTENT\": \"true\"\n }\n }\n}\n```\n\n### Variables de entorno\n\nVariables ambientales clave:\n\n```bash\n# API Keys\nANTHROPIC_API_KEY=anthropic_key\n\n# Chrome Configuration\n# Optional: Path to Chrome executable\nCHROME_PATH=/path/to/chrome\n# Optional: Chrome user data directory\nCHROME_USER_DATA=/path/to/user/data\n# Default: 9222\nCHROME_DEBUGGING_PORT=9222\n# Default: localhost\nCHROME_DEBUGGING_HOST=localhost\n# Keep browser open between tasks\nCHROME_PERSISTENT_SESSION=false\n\n# Model Settings\n# Options: anthropic, openai, azure, deepseek\nMCP_MODEL_PROVIDER=anthropic\n# Model name\nMCP_MODEL_NAME=claude-3-5-sonnet-20241022\nMCP_TEMPERATURE=0.3\nMCP_MAX_STEPS=30\nMCP_USE_VISION=true\nMCP_MAX_ACTIONS_PER_STEP=5\n```\n\n## Desarrollo\n\n### Configuración\n\n1. Clonar el repositorio:\n\n```bash\ngit clone https://github.com/JovaniPink/mcp-browser-use.git\ncd mcp-browser-use\n```\n\n2. Crear y activar entorno virtual:\n\n```bash\npython -m venv .venv\nsource .venv/bin/activate # On Windows: .venv\\Scripts\\activate\n```\n\n3. Instalar dependencias:\n\n```bash\nuv sync\n```\n\n4. Iniciar el servidor\n\n```bash\nuv run mcp-browser-use\n```\n\n### Depuración\n\nPara depurar, utilice el [Inspector MCP](https://github.com/modelcontextprotocol/inspector) :\n\n```bash\nnpx @modelcontextprotocol/inspector uv --directory /path/to/project run mcp-server-browser-use\n```\n\nEl Inspector mostrará una URL para la interfaz de depuración.\n\n## Acciones del navegador\n\nEl servidor admite varias acciones del navegador a través del lenguaje natural:\n\n* Navegación: Ir a URL, retroceder/avanzar, actualizar\n* Interacción: hacer clic, escribir, desplazarse, pasar el cursor\n* Formularios: Complete formularios, envíelos, seleccione opciones\n* Estado: Obtener el contenido de la página, tomar capturas de pantalla\n* Pestañas: crear, cerrar, cambiar entre pestañas\n* Visión: Encuentra elementos por apariencia visual\n* Cookies y almacenamiento: Administrar el estado del navegador\n\n## Seguridad\n\nCabe señalar que existen algunas configuraciones de Chrome que permiten que el servidor controle el navegador. Esto representa un riesgo de seguridad y debe usarse con precaución. El servidor no está diseñado para usarse en un entorno de producción.\n\nDetalles de seguridad: [SECURITY.MD](./documentation/SECURITY.md)\n\n## Contribuyendo\n\nAgradecemos las contribuciones a este proyecto. Siga estos pasos:\n\n1. Bifurcar este repositorio.\n2. Crea tu rama de características: `git checkout -b my-new-feature` .\n3. Confirme sus cambios: `git commit -m 'Add some feature'` .\n4. Empujar a la rama: `git push origin my-new-feature` .\n5. Enviar una solicitud de extracción.\n\nPara cambios importantes, primero abra una incidencia para comentar qué desea cambiar. Actualice las pruebas según corresponda para reflejar los cambios realizados.","Spanish","es-ES",{"_29":56,"_37":60,"_39":61},{"_31":57,"_33":58,"_35":59},"AI エージェントが自然言語を使用して Web ブラウザーと対話できるようにします。自動ブラウジング、フォーム入力、ビジョンベースの要素検出、構造化された JSON 応答による体系的なブラウザー制御などの機能を備えています。","MCPブラウザ使用サーバー","# ブラウザ使用のMCPサーバー\n\n[![鍛冶屋のバッジ](https://smithery.ai/badge/@JovaniPink/mcp-browser-use)](https://smithery.ai/server/@JovaniPink/mcp-browser-use)\n\n> [ブラウザ用](https://github.com/browser-use/browser-use)のMCPサーバー。\n\n## 概要\n\nこのリポジトリには、[ブラウザ用ライブラリ](https://github.com/browser-use/browser-use)用のサーバーが含まれています。このサーバーは、AIエージェントが自然言語を介してWebブラウザと対話できるようにする強力なブラウザ自動化システムを提供します。このサーバーはAnthropicの[モデルコンテキストプロトコル（MCP）](https://modelcontextprotocol.io/introduction)に基づいて構築されており、[ブラウザ用](https://github.com/browser-use/browser-use)ライブラリとのシームレスな統合を提供します。\n\n## 特徴\n\n1. **ブラウザコントロール**\n\n* 自然言語によるブラウザインタラクションの自動化\n* ナビゲーション、フォーム入力、クリック、スクロール機能\n* タブ管理とスクリーンショット機能\n* クッキーと状態の管理\n\n2. **エージェントシステム**\n\n* custom\\_agent.py でのカスタムエージェントの実装\n* 視覚ベースの要素検出\n* アクションに対する構造化されたJSONレスポンス\n* メッセージ履歴の管理と要約\n\n3. **構成**\n\n* APIキーと設定の環境ベースの構成\n* Chromeブラウザの設定（デバッグポート、永続性）\n* モデルプロバイダーの選択とパラメータ\n\n## 依存関係\n\nこのプロジェクトは次の Python パッケージに依存しています。\n\n| パッケージ | バージョン | 説明 |\n| :---------------------------------------------------------------------------------------------------- | :-------- | :------------------------------------------------------------------------------------------------------------------- |\n| [枕](https://python-pillow.org/) | >=10.1.0 | Python インタープリターに画像処理機能を追加する Python Imaging Library (PIL) フォーク。 |\n| [ブラウザの使用](https://github.com/browser-use/browser-use) | ==0.1.19 | AIエージェントが自然言語でウェブブラウザと対話できるようにする強力なブラウザ自動化システム。このプロジェクトのブラウザ自動化機能を支えるコアライブラリです。 |\n| [ファスタピ](https://fastapi.tiangolo.com/) | >=0.115.6 | Python 3.7以降で標準のPython型ヒントに基づいてAPIを構築するための、最新の高速（高パフォーマンス）ウェブフレームワークです。エージェントの機能を公開するサーバーの作成に使用されます。 |\n| [ファストmcp](https://pypi.org/project/fastmcp/) | >=0.4.1 | MCP (Model Context Protocol) サーバーを構築するための FastAPI をラップするフレームワーク。 |\n| [インストラクター](https://github.com/jxnl/instructor) | >=1.7.2 | OpenAIモデルを用いた構造化出力のプロンプトと検証のためのライブラリ。モデルの応答から構造化データを抽出できます。 |\n| [ランチェーン](https://www.langchain.com/) | >=0.3.14 | 大規模言語モデル（LLM）を用いたアプリケーション開発のためのフレームワーク。異なる言語モデルコンポーネントを連結し、様々なAPIやデータソースと連携するためのツールを提供します。 |\n| [langchain-google-genai](https://pypi.org/project/langchain-google-genai/) | >=2.1.1 | Google GenAI モデル用の LangChain 統合により、LangChain フレームワーク内で Google の生成 AI 機能を使用できるようになります。 |\n| [langchain-openai](https://api.python.langchain.com/en/latest/langchain_openai.html) | >=0.2.14 | LangChainとOpenAIのモデルとの統合。LangChainフレームワーク内でOpenAIモデル（GPT-4など）の使用を可能にします。本プロジェクトでは、OpenAIの言語モデルおよびビジョンモデルとの連携に使用されます。 |\n| [ランチェイン・オラマ](https://api.python.langchain.com/en/latest/langchain_ollama/chat_models/ChatOllama.html) | >=0.2.2 | Ollama の Langchain 統合により、LLM のローカル実行が可能になります。 |\n| [オープンアイ](https://platform.openai.com/docs/api-reference) | >=1.59.5 | OpenAI APIの公式Pythonクライアントライブラリ。OpenAIのモデルと直接やり取りするために使用します（必要に応じてLangChainに加えて）。 |\n| [Python-dotenv](https://pypi.org/project/python-dotenv/) | >=1.0.1 | `.env`ファイルからキーと値のペアを読み取り、環境変数として設定します。ローカル開発と構成管理を簡素化します。 |\n| [ピダンティック](https://docs.pydantic.dev/) | >=2.10.5 | Python型アノテーションを用いたデータ検証と設定管理。実行時の型強制と自動モデル作成機能を提供します。エージェント内で構造化データモデルを定義するために不可欠です。 |\n| [パイパークリップ](https://pyperclip.readthedocs.io/) | 1.9.0以上 | クリップボードのコピーと貼り付け機能用のクロスプラットフォーム Python モジュール。 |\n| [ウビコーン](https://www.uvicorn.org/) | >=0.22.0 | Python用のASGIウェブサーバー実装。FastAPIアプリケーションを提供するために使用されます。 |\n\n## コンポーネント\n\n### リソース\n\nサーバーは、次のブラウザ自動化システムを実装します。\n\n* 高度なブラウザ制御のためのブラウザ使用ライブラリとの統合\n* カスタムブラウザ自動化機能\n* 視覚機能を備えたエージェントベースのインタラクションシステム\n* 永続的な状態管理\n* カスタマイズ可能なモデル設定\n\n### 要件\n\n* オペレーティングシステム (Linux、macOS、Windows。Docker または Microsoft WSL についてはテストしていません)\n* Python 3.11以上\n* uv (高速 Python パッケージインストーラー)\n* Chrome/Chromiumブラウザ\n* [クロードデスクトップ](https://claude.ai/download)\n\n### クイックスタート\n\n#### クロードデスクトップ\n\nMacOS の場合: `~/Library/Application\\ Support/Claude/claude_desktop_config.json` Windows の場合: `%APPDATA%/Claude/claude_desktop_config.json`\n\n#### Smithery経由でインストール\n\n[Smithery](https://smithery.ai/server/@JovaniPink/mcp-browser-use)経由で Claude Desktop の Browser Use を自動的にインストールするには:\n\n```bash\nnpx -y @smithery/cli install @JovaniPink/mcp-browser-use --client claude\n```\n\n```json\n\"mcpServers\": {\n \"mcp_server_browser_use\": {\n \"command\": \"uvx\",\n \"args\": [\n \"mcp-server-browser-use\",\n ],\n \"env\": {\n \"OPENAI_ENDPOINT\": \"https://api.openai.com/v1\",\n \"OPENAI_API_KEY\": \"\",\n \"ANTHROPIC_API_KEY\": \"\",\n \"GOOGLE_API_KEY\": \"\",\n \"AZURE_OPENAI_ENDPOINT\": \"\",\n \"AZURE_OPENAI_API_KEY\": \"\",\n // \"DEEPSEEK_ENDPOINT\": \"https://api.deepseek.com\",\n // \"DEEPSEEK_API_KEY\": \"\",\n // Set to false to disable anonymized telemetry\n \"ANONYMIZED_TELEMETRY\": \"false\",\n // Chrome settings\n \"CHROME_PATH\": \"\",\n \"CHROME_USER_DATA\": \"\",\n \"CHROME_DEBUGGING_PORT\": \"9222\",\n \"CHROME_DEBUGGING_HOST\": \"localhost\",\n // Set to true to keep browser open between AI tasks\n \"CHROME_PERSISTENT_SESSION\": \"false\",\n // Model settings\n \"MCP_MODEL_PROVIDER\": \"anthropic\",\n \"MCP_MODEL_NAME\": \"claude-3-5-sonnet-20241022\",\n \"MCP_TEMPERATURE\": \"0.3\",\n \"MCP_MAX_STEPS\": \"30\",\n \"MCP_USE_VISION\": \"true\",\n \"MCP_MAX_ACTIONS_PER_STEP\": \"5\",\n \"MCP_TOOL_CALL_IN_CONTENT\": \"true\"\n }\n }\n}\n```\n\n### 環境変数\n\n主要な環境変数:\n\n```bash\n# API Keys\nANTHROPIC_API_KEY=anthropic_key\n\n# Chrome Configuration\n# Optional: Path to Chrome executable\nCHROME_PATH=/path/to/chrome\n# Optional: Chrome user data directory\nCHROME_USER_DATA=/path/to/user/data\n# Default: 9222\nCHROME_DEBUGGING_PORT=9222\n# Default: localhost\nCHROME_DEBUGGING_HOST=localhost\n# Keep browser open between tasks\nCHROME_PERSISTENT_SESSION=false\n\n# Model Settings\n# Options: anthropic, openai, azure, deepseek\nMCP_MODEL_PROVIDER=anthropic\n# Model name\nMCP_MODEL_NAME=claude-3-5-sonnet-20241022\nMCP_TEMPERATURE=0.3\nMCP_MAX_STEPS=30\nMCP_USE_VISION=true\nMCP_MAX_ACTIONS_PER_STEP=5\n```\n\n## 発達\n\n### 設定\n\n1. リポジトリをクローンします。\n\n```bash\ngit clone https://github.com/JovaniPink/mcp-browser-use.git\ncd mcp-browser-use\n```\n\n2. 仮想環境を作成してアクティブ化します。\n\n```bash\npython -m venv .venv\nsource .venv/bin/activate # On Windows: .venv\\Scripts\\activate\n```\n\n3. 依存関係をインストールします:\n\n```bash\nuv sync\n```\n\n4. サーバーを起動する\n\n```bash\nuv run mcp-browser-use\n```\n\n### デバッグ\n\nデバッグには、 [MCP Inspector](https://github.com/modelcontextprotocol/inspector)を使用します。\n\n```bash\nnpx @modelcontextprotocol/inspector uv --directory /path/to/project run mcp-server-browser-use\n```\n\nインスペクターにはデバッグインターフェイスの URL が表示されます。\n\n## ブラウザアクション\n\nサーバーは自然言語によるさまざまなブラウザアクションをサポートします。\n\n* ナビゲーション: URL に移動、戻る/進む、更新\n* インタラクション: クリック、入力、スクロール、ホバー\n* フォーム: フォームに入力し、送信し、オプションを選択します\n* 状態: ページコンテンツを取得し、スクリーンショットを撮る\n* タブ: タブの作成、閉じる、タブ間の切り替え\n* 視覚: 視覚的な外観で要素を見つける\n* クッキーとストレージ: ブラウザの状態を管理する\n\n## 安全\n\nChromeの設定によっては、ブラウザをサーバー側で制御できるように設定されているものがあることにご注意ください。これはセキュリティリスクとなるため、注意して使用してください。このサーバーは本番環境での使用を想定していません。\n\nセキュリティの詳細: [SECURITY.MD](./documentation/SECURITY.md)\n\n## 貢献\n\nこのプロジェクトへの貢献を歓迎します。以下の手順に従ってください。\n\n1. このリポジトリをフォークします。\n2. 機能ブランチを作成します: `git checkout -b my-new-feature` 。\n3. 変更をコミットします: `git commit -m 'Add some feature'` 。\n4. ブランチにプッシュします: `git push origin my-new-feature` 。\n5. プルリクエストを送信します。\n\n大きな変更については、まず問題を開いて変更内容についてご相談ください。変更内容を反映し、必要に応じてテストを更新してください。","Japanese","ja-JP",{"_29":63,"_37":67,"_39":68},{"_31":64,"_33":65,"_35":66},"AI 에이전트가 자연어를 사용하여 웹 브라우저와 상호 작용할 수 있도록 하며, 자동 브라우징, 양식 작성, 비전 기반 요소 감지, 체계적인 브라우저 제어를 위한 구조화된 JSON 응답 기능을 제공합니다.","MCP 브라우저 사용 서버","# 브라우저 사용이 가능한 MCP 서버\n\n[![대장간 배지](https://smithery.ai/badge/@JovaniPink/mcp-browser-use)](https://smithery.ai/server/@JovaniPink/mcp-browser-use)\n\n> [브라우저에서 사용할 수](https://github.com/browser-use/browser-use) 있는 MCP 서버입니다.\n\n## 개요\n\n이 저장소에는 [브라우저 기반](https://github.com/browser-use/browser-use) 라이브러리 서버가 포함되어 있으며, 이 라이브러리는 AI 에이전트가 자연어를 통해 웹 브라우저와 상호 작용할 수 있도록 하는 강력한 브라우저 자동화 시스템을 제공합니다. 이 서버는 Anthropic의 [모델 컨텍스트 프로토콜(MCP)을](https://modelcontextprotocol.io/introduction) 기반으로 구축되었으며 [브라우저 기반](https://github.com/browser-use/browser-use) 라이브러리와 완벽하게 통합됩니다.\n\n## 특징\n\n1. **브라우저 컨트롤**\n\n* 자연어를 통한 자동화된 브라우저 상호 작용\n* 탐색, 양식 작성, 클릭 및 스크롤 기능\n* 탭 관리 및 스크린샷 기능\n* 쿠키 및 상태 관리\n\n2. **에이전트 시스템**\n\n* custom\\_agent.py에서 사용자 정의 에이전트 구현\n* 비전 기반 요소 감지\n* 작업에 대한 구조화된 JSON 응답\n* 메시지 기록 관리 및 요약\n\n3. **구성**\n\n* API 키 및 설정에 대한 환경 기반 구성\n* Chrome 브라우저 설정(디버깅 포트, 지속성)\n* 모델 제공자 선택 및 매개변수\n\n## 종속성\n\n이 프로젝트는 다음과 같은 Python 패키지를 사용합니다.\n\n| 패키지 | 버전 | 설명 |\n| :------------------------------------------------------------------------------------------------- | :-------- | :--------------------------------------------------------------------------------------------------------------------------------- |\n| [베개](https://python-pillow.org/) | >=10.1.0 | Python 인터프리터에 이미지 처리 기능을 추가하는 Python 이미징 라이브러리(PIL) 포크입니다. |\n| [브라우저 사용](https://github.com/browser-use/browser-use) | ==0.1.19 | AI 에이전트가 자연어를 통해 웹 브라우저와 상호 작용할 수 있도록 지원하는 강력한 브라우저 자동화 시스템입니다. 이 프로젝트의 브라우저 자동화 기능을 구동하는 핵심 라이브러리입니다. |\n| [패스트아피](https://fastapi.tiangolo.com/) | >=0.115.6 | Python 3.7 이상 버전으로 표준 Python 타입 힌트를 기반으로 API를 구축하기 위한 최신의 빠른(고성능) 웹 프레임워크입니다. 에이전트 기능을 제공하는 서버를 생성하는 데 사용됩니다. |\n| [패스트MCP](https://pypi.org/project/fastmcp/) | >=0.4.1 | MCP(Model Context Protocol) 서버를 구축하기 위한 FastAPI를 래핑하는 프레임워크입니다. |\n| [강사](https://github.com/jxnl/instructor) | >=1.7.2 | OpenAI 모델을 사용하여 구조화된 출력 프롬프트 및 검증을 위한 라이브러리입니다. 모델 응답에서 구조화된 데이터를 추출할 수 있습니다. |\n| [랭체인](https://www.langchain.com/) | >=0.3.14 | 대규모 언어 모델(LLM)을 사용하는 애플리케이션을 개발하기 위한 프레임워크입니다. 서로 다른 언어 모델 구성 요소를 연결하고 다양한 API 및 데이터 소스와 상호 작용할 수 있는 도구를 제공합니다. |\n| [langchain-google-genai](https://pypi.org/project/langchain-google-genai/) | >=2.1.1 | Google GenAI 모델에 LangChain을 통합하여 LangChain 프레임워크 내에서 Google의 생성 AI 기능을 사용할 수 있습니다. |\n| [랭체인-오픈AI](https://api.python.langchain.com/en/latest/langchain_openai.html) | >=0.2.14 | LangChain과 OpenAI 모델이 통합되었습니다. LangChain 프레임워크 내에서 OpenAI 모델(예: GPT-4)을 사용할 수 있습니다. 이 프로젝트에서는 OpenAI의 언어 및 비전 모델과 상호 작용하는 데 사용됩니다. |\n| [랭체인-올라마](https://api.python.langchain.com/en/latest/langchain_ollama/chat_models/ChatOllama.html) | >=0.2.2 | Ollama에 Langchain을 통합하여 LLM의 로컬 실행을 지원합니다. |\n| [오픈AI](https://platform.openai.com/docs/api-reference) | >=1.59.5 | OpenAI API용 공식 Python 클라이언트 라이브러리입니다. OpenAI 모델과 직접 상호 작용하는 데 사용됩니다(필요한 경우 LangChain 외에도 사용 가능). |\n| [파이썬-dotenv](https://pypi.org/project/python-dotenv/) | >=1.0.1 | `.env` 파일에서 키-값 쌍을 읽어 환경 변수로 설정합니다. 로컬 개발 및 구성 관리를 간소화합니다. |\n| [피단틱](https://docs.pydantic.dev/) | >=2.10.5 | Python 타입 어노테이션을 사용한 데이터 검증 및 설정 관리. 런타임 시 타입 적용 및 자동 모델 생성 기능을 제공합니다. 에이전트에서 구조화된 데이터 모델을 정의하는 데 필수적입니다. |\n| [파이퍼클립](https://pyperclip.readthedocs.io/) | >=1.9.0 | 클립보드 기능을 복사하고 붙여넣기 위한 크로스 플랫폼 Python 모듈입니다. |\n| [유비콘](https://www.uvicorn.org/) | >=0.22.0 | Python용 ASGI 웹 서버 구현. FastAPI 애플리케이션 제공에 사용됩니다. |\n\n## 구성 요소\n\n### 자원\n\n서버는 다음을 사용하여 브라우저 자동화 시스템을 구현합니다.\n\n* 고급 브라우저 제어를 위한 브라우저 사용 라이브러리와의 통합\n* 사용자 정의 브라우저 자동화 기능\n* 비전 기능을 갖춘 에이전트 기반 상호작용 시스템\n* 지속적인 상태 관리\n* 사용자 정의 가능한 모델 설정\n\n### 요구 사항\n\n* 운영 체제(Linux, macOS, Windows; Docker 또는 Microsoft WSL은 테스트하지 않았습니다)\n* Python 3.11 이상\n* uv(빠른 Python 패키지 설치 프로그램)\n* 크롬/크로미엄 브라우저\n* [클로드 데스크탑](https://claude.ai/download)\n\n### 빠른 시작\n\n#### 클로드 데스크탑\n\nMacOS의 경우: `~/Library/Application\\ Support/Claude/claude_desktop_config.json` Windows의 경우: `%APPDATA%/Claude/claude_desktop_config.json`\n\n#### Smithery를 통해 설치\n\n[Smithery를](https://smithery.ai/server/@JovaniPink/mcp-browser-use) 통해 Claude Desktop용 Browser Use를 자동으로 설치하려면:\n\n지엑스피1\n\n```json\n\"mcpServers\": {\n \"mcp_server_browser_use\": {\n \"command\": \"uvx\",\n \"args\": [\n \"mcp-server-browser-use\",\n ],\n \"env\": {\n \"OPENAI_ENDPOINT\": \"https://api.openai.com/v1\",\n \"OPENAI_API_KEY\": \"\",\n \"ANTHROPIC_API_KEY\": \"\",\n \"GOOGLE_API_KEY\": \"\",\n \"AZURE_OPENAI_ENDPOINT\": \"\",\n \"AZURE_OPENAI_API_KEY\": \"\",\n // \"DEEPSEEK_ENDPOINT\": \"https://api.deepseek.com\",\n // \"DEEPSEEK_API_KEY\": \"\",\n // Set to false to disable anonymized telemetry\n \"ANONYMIZED_TELEMETRY\": \"false\",\n // Chrome settings\n \"CHROME_PATH\": \"\",\n \"CHROME_USER_DATA\": \"\",\n \"CHROME_DEBUGGING_PORT\": \"9222\",\n \"CHROME_DEBUGGING_HOST\": \"localhost\",\n // Set to true to keep browser open between AI tasks\n \"CHROME_PERSISTENT_SESSION\": \"false\",\n // Model settings\n \"MCP_MODEL_PROVIDER\": \"anthropic\",\n \"MCP_MODEL_NAME\": \"claude-3-5-sonnet-20241022\",\n \"MCP_TEMPERATURE\": \"0.3\",\n \"MCP_MAX_STEPS\": \"30\",\n \"MCP_USE_VISION\": \"true\",\n \"MCP_MAX_ACTIONS_PER_STEP\": \"5\",\n \"MCP_TOOL_CALL_IN_CONTENT\": \"true\"\n }\n }\n}\n```\n\n### 환경 변수\n\n주요 환경 변수:\n\n```bash\n# API Keys\nANTHROPIC_API_KEY=anthropic_key\n\n# Chrome Configuration\n# Optional: Path to Chrome executable\nCHROME_PATH=/path/to/chrome\n# Optional: Chrome user data directory\nCHROME_USER_DATA=/path/to/user/data\n# Default: 9222\nCHROME_DEBUGGING_PORT=9222\n# Default: localhost\nCHROME_DEBUGGING_HOST=localhost\n# Keep browser open between tasks\nCHROME_PERSISTENT_SESSION=false\n\n# Model Settings\n# Options: anthropic, openai, azure, deepseek\nMCP_MODEL_PROVIDER=anthropic\n# Model name\nMCP_MODEL_NAME=claude-3-5-sonnet-20241022\nMCP_TEMPERATURE=0.3\nMCP_MAX_STEPS=30\nMCP_USE_VISION=true\nMCP_MAX_ACTIONS_PER_STEP=5\n```\n\n## 개발\n\n### 설정\n\n1. 저장소를 복제합니다.\n\n```bash\ngit clone https://github.com/JovaniPink/mcp-browser-use.git\ncd mcp-browser-use\n```\n\n2. 가상 환경을 만들고 활성화하세요.\n\n```bash\npython -m venv .venv\nsource .venv/bin/activate # On Windows: .venv\\Scripts\\activate\n```\n\n3. 종속성 설치:\n\n```bash\nuv sync\n```\n\n4. 서버를 시작합니다\n\n```bash\nuv run mcp-browser-use\n```\n\n### 디버깅\n\n디버깅을 위해 [MCP Inspector를](https://github.com/modelcontextprotocol/inspector) 사용하세요.\n\n```bash\nnpx @modelcontextprotocol/inspector uv --directory /path/to/project run mcp-server-browser-use\n```\n\n검사기는 디버깅 인터페이스의 URL을 표시합니다.\n\n## 브라우저 동작\n\n서버는 자연어를 통해 다양한 브라우저 동작을 지원합니다.\n\n* 탐색: URL로 이동, 뒤로/앞으로, 새로 고침\n* 상호작용: 클릭, 입력, 스크롤, 호버\n* 양식: 양식 작성, 제출, 옵션 선택\n* 상태: 페이지 콘텐츠 가져오기, 스크린샷 찍기\n* 탭: 탭 생성, 닫기, 탭 간 전환\n* 비전: 시각적 모양으로 요소 찾기\n* 쿠키 및 저장소: 브라우저 상태 관리\n\n## 보안\n\nChrome 설정 중 일부에서는 브라우저를 서버에서 제어할 수 있도록 설정되어 있습니다. 이는 보안 위험이므로 주의해서 사용해야 합니다. 이 서버는 프로덕션 환경에서 사용하도록 설계되지 않았습니다.\n\n보안 세부 정보: [SECURITY.MD](./documentation/SECURITY.md)\n\n## 기여하다\n\n이 프로젝트에 대한 여러분의 참여를 환영합니다. 다음 단계를 따라주세요.\n\n1. 이 저장소를 포크하세요.\n2. 기능 브랜치를 만듭니다: `git checkout -b my-new-feature` .\n3. 변경 사항을 커밋합니다: `git commit -m 'Add some feature'` .\n4. 브랜치에 푸시합니다: `git push origin my-new-feature` .\n5. 풀 리퀘스트를 제출하세요.\n\n주요 변경 사항의 경우, 먼저 이슈를 열어 변경 사항을 논의하세요. 변경 사항이 반영되도록 테스트를 적절히 업데이트해 주세요.","Korean","ko-KR",{"_29":70,"_37":74,"_39":75},{"_31":71,"_33":72,"_35":73},"Позволяет агентам ИИ взаимодействовать с веб-браузерами, используя естественный язык, обеспечивая автоматизированный просмотр, заполнение форм, обнаружение элементов на основе зрения и структурированные ответы JSON для систематического управления браузером.","Браузер MCP Использовать сервер","# MCP-сервер с использованием браузера\n\n[![значок кузнеца](https://smithery.ai/badge/@JovaniPink/mcp-browser-use)](https://smithery.ai/server/@JovaniPink/mcp-browser-use)\n\n> MCP-сервер для [использования в браузере](https://github.com/browser-use/browser-use) .\n\n## Обзор\n\nЭтот репозиторий содержит сервер для библиотеки [браузера](https://github.com/browser-use/browser-use) , которая обеспечивает мощную систему автоматизации браузера, позволяющую агентам ИИ взаимодействовать с веб-браузерами посредством естественного языка. Сервер построен на [протоколе контекста модели (MCP)](https://modelcontextprotocol.io/introduction) Anthropic и обеспечивает бесшовную интеграцию с библиотекой [браузера](https://github.com/browser-use/browser-use) .\n\n## Функции\n\n1. **Управление браузером**\n\n* Автоматизированное взаимодействие с браузером посредством естественного языка\n* Возможности навигации, заполнения форм, нажатия и прокрутки\n* Управление вкладками и функциональность скриншотов\n* Управление файлами cookie и состоянием\n\n2. **Система Агентов**\n\n* Реализация пользовательского агента в custom\\_agent.py\n* Обнаружение элементов на основе зрения\n* Структурированные ответы JSON для действий\n* Управление историей сообщений и ее обобщение\n\n3. **Конфигурация**\n\n* Конфигурация API-ключей и настроек на основе среды\n* Настройки браузера Chrome (порт отладки, сохранение)\n* Выбор поставщика модели и параметры\n\n## Зависимости\n\nЭтот проект использует следующие пакеты Python:\n\n| Упаковка | Версия | Описание |\n| :---------------------------------------------------------------------------------------------------------- | :-------- | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| [Подушка](https://python-pillow.org/) | >=10.1.0 | Форк библиотеки Python Imaging Library (PIL), добавляющий возможности обработки изображений в ваш интерпретатор Python. |\n| [использование браузера](https://github.com/browser-use/browser-use) | ==0.1.19 | Мощная система автоматизации браузера, которая позволяет агентам ИИ взаимодействовать с веб-браузерами посредством естественного языка. Основная библиотека, которая обеспечивает возможности автоматизации браузера этого проекта. |\n| [фастапи](https://fastapi.tiangolo.com/) | >=0,115,6 | Современный, быстрый (высокопроизводительный) веб-фреймворк для создания API с Python 3.7+ на основе стандартных подсказок типов Python. Используется для создания сервера, который раскрывает функциональность агента. |\n| [fastmcp](https://pypi.org/project/fastmcp/) | >=0.4.1 | Фреймворк, оборачивающий FastAPI для создания серверов MCP (Model Context Protocol). |\n| [инструктор](https://github.com/jxnl/instructor) | >=1.7.2 | Библиотека для структурированного вывода подсказок и проверки с моделями OpenAI. Позволяет извлекать структурированные данные из ответов модели. |\n| [langchain](https://www.langchain.com/) | >=0.3.14 | Фреймворк для разработки приложений с большими языковыми моделями (LLM). Предоставляет инструменты для объединения различных компонентов языковой модели и взаимодействия с различными API и источниками данных. |\n| [langchain-google-genai](https://pypi.org/project/langchain-google-genai/) | >=2.1.1 | Интеграция LangChain для моделей Google GenAI, позволяющая использовать возможности генеративного ИИ Google в рамках LangChain. |\n| [langchain-openai](https://api.python.langchain.com/en/latest/langchain_openai.html) | >=0.2.14 | Интеграция LangChain с моделями OpenAI. Позволяет использовать модели OpenAI (например, GPT-4) в рамках LangChain. Используется в этом проекте для взаимодействия с моделями языка и зрения OpenAI. |\n| [langchain-ollama](https://api.python.langchain.com/en/latest/langchain_ollama/chat_models/ChatOllama.html) | >=0.2.2 | Интеграция Langchain для Ollama, обеспечивающая локальное выполнение LLM. |\n| [опенай](https://platform.openai.com/docs/api-reference) | >=1.59.5 | Официальная клиентская библиотека Python для API OpenAI. Используется для прямого взаимодействия с моделями OpenAI (при необходимости, в дополнение к LangChain). |\n| [python-dotenv](https://pypi.org/project/python-dotenv/) | >=1.0.1 | Считывает пары ключ-значение из файла `.env` и устанавливает их как переменные среды. Упрощает локальную разработку и управление конфигурацией. |\n| [пидантический](https://docs.pydantic.dev/) | >=2.10.5 | Проверка данных и управление настройками с использованием аннотаций типов Python. Обеспечивает принудительное применение типов во время выполнения и автоматическое создание моделей. Необходим для определения структурированных моделей данных в агенте. |\n| [pyperclip](https://pyperclip.readthedocs.io/) | >=1.9.0 | Кроссплатформенный модуль Python для функций копирования и вставки буфера обмена. |\n| [увикорн](https://www.uvicorn.org/) | >=0,22,0 | Реализация веб-сервера ASGI для Python. Используется для обслуживания приложения FastAPI. |\n\n## Компоненты\n\n### Ресурсы\n\nНа сервере реализована система автоматизации браузера с:\n\n* Интеграция с библиотекой браузера для расширенного управления браузером\n* Возможности автоматизации пользовательского браузера\n* Система взаимодействия на основе агентов с возможностями визуализации\n* Постоянное управление состоянием\n* Настраиваемые параметры модели\n\n### Требования\n\n* Операционные системы (Linux, macOS, Windows; мы не тестировали Docker или Microsoft WSL)\n* Python 3.11 или выше\n* uv (быстрый установщик пакетов Python)\n* Браузер Chrome/Chromium\n* [Клод Десктоп](https://claude.ai/download)\n\n### Быстрый старт\n\n#### Клод Десктоп\n\nВ MacOS: `~/Library/Application\\ Support/Claude/claude_desktop_config.json` В Windows: `%APPDATA%/Claude/claude_desktop_config.json`\n\n#### Установка через Smithery\n\nДля автоматической установки Browser Use for Claude Desktop через [Smithery](https://smithery.ai/server/@JovaniPink/mcp-browser-use) :\n\n```bash\nnpx -y @smithery/cli install @JovaniPink/mcp-browser-use --client claude\n```\n\n```json\n\"mcpServers\": {\n \"mcp_server_browser_use\": {\n \"command\": \"uvx\",\n \"args\": [\n \"mcp-server-browser-use\",\n ],\n \"env\": {\n \"OPENAI_ENDPOINT\": \"https://api.openai.com/v1\",\n \"OPENAI_API_KEY\": \"\",\n \"ANTHROPIC_API_KEY\": \"\",\n \"GOOGLE_API_KEY\": \"\",\n \"AZURE_OPENAI_ENDPOINT\": \"\",\n \"AZURE_OPENAI_API_KEY\": \"\",\n // \"DEEPSEEK_ENDPOINT\": \"https://api.deepseek.com\",\n // \"DEEPSEEK_API_KEY\": \"\",\n // Set to false to disable anonymized telemetry\n \"ANONYMIZED_TELEMETRY\": \"false\",\n // Chrome settings\n \"CHROME_PATH\": \"\",\n \"CHROME_USER_DATA\": \"\",\n \"CHROME_DEBUGGING_PORT\": \"9222\",\n \"CHROME_DEBUGGING_HOST\": \"localhost\",\n // Set to true to keep browser open between AI tasks\n \"CHROME_PERSISTENT_SESSION\": \"false\",\n // Model settings\n \"MCP_MODEL_PROVIDER\": \"anthropic\",\n \"MCP_MODEL_NAME\": \"claude-3-5-sonnet-20241022\",\n \"MCP_TEMPERATURE\": \"0.3\",\n \"MCP_MAX_STEPS\": \"30\",\n \"MCP_USE_VISION\": \"true\",\n \"MCP_MAX_ACTIONS_PER_STEP\": \"5\",\n \"MCP_TOOL_CALL_IN_CONTENT\": \"true\"\n }\n }\n}\n```\n\n### Переменные среды\n\nКлючевые переменные среды:\n\n```bash\n# API Keys\nANTHROPIC_API_KEY=anthropic_key\n\n# Chrome Configuration\n# Optional: Path to Chrome executable\nCHROME_PATH=/path/to/chrome\n# Optional: Chrome user data directory\nCHROME_USER_DATA=/path/to/user/data\n# Default: 9222\nCHROME_DEBUGGING_PORT=9222\n# Default: localhost\nCHROME_DEBUGGING_HOST=localhost\n# Keep browser open between tasks\nCHROME_PERSISTENT_SESSION=false\n\n# Model Settings\n# Options: anthropic, openai, azure, deepseek\nMCP_MODEL_PROVIDER=anthropic\n# Model name\nMCP_MODEL_NAME=claude-3-5-sonnet-20241022\nMCP_TEMPERATURE=0.3\nMCP_MAX_STEPS=30\nMCP_USE_VISION=true\nMCP_MAX_ACTIONS_PER_STEP=5\n```\n\n## Разработка\n\n### Настраивать\n\n1. Клонируйте репозиторий:\n\n```bash\ngit clone https://github.com/JovaniPink/mcp-browser-use.git\ncd mcp-browser-use\n```\n\n2. Создать и активировать виртуальную среду:\n\n```bash\npython -m venv .venv\nsource .venv/bin/activate # On Windows: .venv\\Scripts\\activate\n```\n\n3. Установить зависимости:\n\n```bash\nuv sync\n```\n\n4. Запустить сервер\n\n```bash\nuv run mcp-browser-use\n```\n\n### Отладка\n\nДля отладки используйте [MCP Inspector](https://github.com/modelcontextprotocol/inspector) :\n\n```bash\nnpx @modelcontextprotocol/inspector uv --directory /path/to/project run mcp-server-browser-use\n```\n\nИнспектор отобразит URL-адрес интерфейса отладки.\n\n## Действия браузера\n\nСервер поддерживает различные действия браузера посредством естественного языка:\n\n* Навигация: переход по URL-адресам, назад/вперед, обновление\n* Взаимодействие: щелчок, ввод текста, прокрутка, наведение курсора.\n* Формы: заполните формы, отправьте, выберите параметры\n* Состояние: Получить содержимое страницы, сделать скриншоты\n* Вкладки: создание, закрытие, переключение между вкладками\n* Видение: поиск элементов по внешнему виду\n* Файлы cookie и хранилище: управление состоянием браузера\n\n## Безопасность\n\nХочу отметить, что некоторые настройки Chrome настроены так, чтобы сервер мог управлять браузером. Это риск для безопасности и должно использоваться с осторожностью. Сервер не предназначен для использования в производственной среде.\n\nПодробности безопасности: [SECURITY.MD](./documentation/SECURITY.md)\n\n## Внося вклад\n\nМы приветствуем вклады в этот проект. Пожалуйста, выполните следующие шаги:\n\n1. Сделайте форк этого репозитория.\n2. Создайте ветку функций: `git checkout -b my-new-feature` .\n3. Зафиксируйте изменения: `git commit -m 'Add some feature'` .\n4. Отправка в ветку: `git push origin my-new-feature` .\n5. Отправьте запрос на извлечение.\n\nДля крупных изменений сначала откройте вопрос, чтобы обсудить, что вы хотели бы изменить. Пожалуйста, обновите тесты соответствующим образом, чтобы отразить любые внесенные изменения.","Russian","ru-RU",{"_29":77,"_37":81,"_39":82},{"_31":78,"_33":79,"_35":80},"使 AI 代理能够使用自然语言与 Web 浏览器交互，具有自动浏览、表单填写、基于视觉的元素检测和结构化 JSON 响应等功能，可实现系统的浏览器控制。","MCP 浏览器使用服务器","# MCP 服务器与浏览器一起使用\n\n[![铁匠徽章](https://smithery.ai/badge/@JovaniPink/mcp-browser-use)](https://smithery.ai/server/@JovaniPink/mcp-browser-use)\n\n> 供[浏览器使用的](https://github.com/browser-use/browser-use)MCP 服务器。\n\n## 概述\n\n此存储库包含[浏览器使用](https://github.com/browser-use/browser-use)库的服务器，该库提供了一个强大的浏览器自动化系统，使AI代理能够通过自然语言与Web浏览器交互。该服务器基于Anthropic的[模型上下文协议（MCP）](https://modelcontextprotocol.io/introduction)构建，并与[浏览器使用](https://github.com/browser-use/browser-use)库无缝集成。\n\n## 特征\n\n1. **浏览器控制**\n\n* 通过自然语言实现浏览器自动交互\n* 导航、表单填写、点击和滚动功能\n* 标签管理和屏幕截图功能\n* Cookie 和状态管理\n\n2. **代理系统**\n\n* custom\\_agent.py 中的自定义代理实现\n* 基于视觉的元素检测\n* 操作的结构化 JSON 响应\n* 消息历史管理和汇总\n\n3. **配置**\n\n* 基于环境的 API 密钥和设置配置\n* Chrome 浏览器设置（调试端口、持久性）\n* 模型提供商选择和参数\n\n## 依赖项\n\n该项目依赖于以下 Python 包：\n\n| 包裹 | 版本 | 描述 |\n| :------------------------------------------------------------------------------------------------- | :-------- | :---------------------------------------------------------------------------------------------- |\n| [枕头](https://python-pillow.org/) | >=10.1.0 | Python 图像库 (PIL) 分支为您的 Python 解释器添加了图像处理功能。 |\n| [浏览器使用](https://github.com/browser-use/browser-use) | ==0.1.19 | 一个强大的浏览器自动化系统，使AI代理能够通过自然语言与Web浏览器交互。该项目浏览器自动化功能的核心库。 |\n| [fastapi](https://fastapi.tiangolo.com/) | >=0.115.6 | 基于标准 Python 类型提示，使用 Python 3.7+ 构建 API 的现代、快速（高性能）Web 框架。用于创建公开代理功能的服务器。 |\n| [fastmcp](https://pypi.org/project/fastmcp/) | >=0.4.1 | 包装 FastAPI 以构建 MCP（模型上下文协议）服务器的框架。 |\n| [讲师](https://github.com/jxnl/instructor) | >=1.7.2 | 用于使用 OpenAI 模型进行结构化输出提示和验证的库。支持从模型响应中提取结构化数据。 |\n| [朗链](https://www.langchain.com/) | >=0.3.14 | 使用大型语言模型 (LLM) 开发应用程序的框架。提供用于链接不同语言模型组件并与各种 API 和数据源交互的工具。 |\n| [langchain-google-genai](https://pypi.org/project/langchain-google-genai/) | >=2.1.1 | LangChain 与 Google GenAI 模型的集成，使得能够在 LangChain 框架内使用 Google 的生成式 AI 功能。 |\n| [langchain-openai](https://api.python.langchain.com/en/latest/langchain_openai.html) | >=0.2.14 | LangChain 与 OpenAI 模型集成。支持在 LangChain 框架内使用 OpenAI 模型（例如 GPT-4）。在本项目中，它用于与 OpenAI 的语言和视觉模型进行交互。 |\n| [朗查因－奥拉马](https://api.python.langchain.com/en/latest/langchain_ollama/chat_models/ChatOllama.html) | >=0.2.2 | Ollama 的 Langchain 集成，支持 LLM 的本地执行。 |\n| [openai](https://platform.openai.com/docs/api-reference) | >=1.59.5 | OpenAI API 的官方 Python 客户端库。用于直接与 OpenAI 的模型交互（如果需要，还可以与 LangChain 交互）。 |\n| [python-dotenv](https://pypi.org/project/python-dotenv/) | >=1.0.1 | 从`.env`文件读取键值对并将其设置为环境变量。简化本地开发和配置管理。 |\n| [派丹蒂克](https://docs.pydantic.dev/) | >=2.10.5 | 使用 Python 类型注释进行数据验证和设置管理。提供运行时类型强制执行和自动模型创建功能。这对于在代理中定义结构化数据模型至关重要。 |\n| [pyperclip](https://pyperclip.readthedocs.io/) | >=1.9.0 | 用于复制和粘贴剪贴板功能的跨平台 Python 模块。 |\n| [uvicorn](https://www.uvicorn.org/) | >=0.22.0 | 适用于 Python 的 ASGI Web 服务器实现。用于服务 FastAPI 应用。 |\n\n## 成分\n\n### 资源\n\n该服务器通过以下方式实现浏览器自动化系统：\n\n* 与浏览器使用库集成以实现高级浏览器控制\n* 自定义浏览器自动化功能\n* 具有视觉功能的基于代理的交互系统\n* 持久状态管理\n* 可定制的模型设置\n\n### 要求\n\n* 操作系统（Linux、macOS、Windows；我们尚未测试 Docker 或 Microsoft WSL）\n* Python 3.11 或更高版本\n* uv（快速 Python 包安装程序）\n* Chrome/Chromium浏览器\n* [克劳德桌面](https://claude.ai/download)\n\n### 快速入门\n\n#### 克劳德桌面\n\n在 MacOS 上： `~/Library/Application\\ Support/Claude/claude_desktop_config.json`在 Windows 上： `%APPDATA%/Claude/claude_desktop_config.json`\n\n#### 通过 Smithery 安装\n\n要通过[Smithery](https://smithery.ai/server/@JovaniPink/mcp-browser-use)自动安装 Claude Desktop 的浏览器使用：\n\n```bash\nnpx -y @smithery/cli install @JovaniPink/mcp-browser-use --client claude\n```\n\n```json\n\"mcpServers\": {\n \"mcp_server_browser_use\": {\n \"command\": \"uvx\",\n \"args\": [\n \"mcp-server-browser-use\",\n ],\n \"env\": {\n \"OPENAI_ENDPOINT\": \"https://api.openai.com/v1\",\n \"OPENAI_API_KEY\": \"\",\n \"ANTHROPIC_API_KEY\": \"\",\n \"GOOGLE_API_KEY\": \"\",\n \"AZURE_OPENAI_ENDPOINT\": \"\",\n \"AZURE_OPENAI_API_KEY\": \"\",\n // \"DEEPSEEK_ENDPOINT\": \"https://api.deepseek.com\",\n // \"DEEPSEEK_API_KEY\": \"\",\n // Set to false to disable anonymized telemetry\n \"ANONYMIZED_TELEMETRY\": \"false\",\n // Chrome settings\n \"CHROME_PATH\": \"\",\n \"CHROME_USER_DATA\": \"\",\n \"CHROME_DEBUGGING_PORT\": \"9222\",\n \"CHROME_DEBUGGING_HOST\": \"localhost\",\n // Set to true to keep browser open between AI tasks\n \"CHROME_PERSISTENT_SESSION\": \"false\",\n // Model settings\n \"MCP_MODEL_PROVIDER\": \"anthropic\",\n \"MCP_MODEL_NAME\": \"claude-3-5-sonnet-20241022\",\n \"MCP_TEMPERATURE\": \"0.3\",\n \"MCP_MAX_STEPS\": \"30\",\n \"MCP_USE_VISION\": \"true\",\n \"MCP_MAX_ACTIONS_PER_STEP\": \"5\",\n \"MCP_TOOL_CALL_IN_CONTENT\": \"true\"\n }\n }\n}\n```\n\n### 环境变量\n\n关键环境变量：\n\n```bash\n# API Keys\nANTHROPIC_API_KEY=anthropic_key\n\n# Chrome Configuration\n# Optional: Path to Chrome executable\nCHROME_PATH=/path/to/chrome\n# Optional: Chrome user data directory\nCHROME_USER_DATA=/path/to/user/data\n# Default: 9222\nCHROME_DEBUGGING_PORT=9222\n# Default: localhost\nCHROME_DEBUGGING_HOST=localhost\n# Keep browser open between tasks\nCHROME_PERSISTENT_SESSION=false\n\n# Model Settings\n# Options: anthropic, openai, azure, deepseek\nMCP_MODEL_PROVIDER=anthropic\n# Model name\nMCP_MODEL_NAME=claude-3-5-sonnet-20241022\nMCP_TEMPERATURE=0.3\nMCP_MAX_STEPS=30\nMCP_USE_VISION=true\nMCP_MAX_ACTIONS_PER_STEP=5\n```\n\n## 发展\n\n### 设置\n\n1. 克隆存储库：\n\n```bash\ngit clone https://github.com/JovaniPink/mcp-browser-use.git\ncd mcp-browser-use\n```\n\n2. 创建并激活虚拟环境：\n\n```bash\npython -m venv .venv\nsource .venv/bin/activate # On Windows: .venv\\Scripts\\activate\n```\n\n3. 安装依赖项：\n\n```bash\nuv sync\n```\n\n4. 启动服务器\n\n```bash\nuv run mcp-browser-use\n```\n\n### 调试\n\n如需调试，请使用[MCP Inspector](https://github.com/modelcontextprotocol/inspector) ：\n\n```bash\nnpx @modelcontextprotocol/inspector uv --directory /path/to/project run mcp-server-browser-use\n```\n\n检查器将显示调试接口的 URL。\n\n## 浏览器操作\n\n服务器通过自然语言支持各种浏览器操作：\n\n* 导航：转到 URL、后退/前进、刷新\n* 交互：点击、输入、滚动、悬停\n* 表格：填写表格、提交、选择选项\n* 状态：获取页面内容，截取屏幕截图\n* 标签：创建、关闭、切换标签\n* 视觉：通过视觉外观查找元素\n* Cookies 和存储：管理浏览器状态\n\n## 安全\n\n需要注意的是，Chrome 的一些设置允许服务器控制浏览器。这存在安全风险，应谨慎使用。该服务器不适用于生产环境。\n\n安全详情： [SECURITY.MD](./documentation/SECURITY.md)\n\n## 贡献\n\n欢迎大家为本项目贡献力量。请按以下步骤操作：\n\n1. 分叉此存储库。\n2. 创建你的功能分支： `git checkout -b my-new-feature` 。\n3. 提交您的更改： `git commit -m 'Add some feature'` 。\n4. 推送到分支： `git push origin my-new-feature` 。\n5. 提交拉取请求。\n\n对于重大变更，请先提交问题讨论您想要修改的内容。请根据需要更新测试以反映所做的任何更改。","Chinese","zh-CN","githubUser","locale",{"_86":87,"_13":88,"_97":43,"_98":43,"_33":44,"_16":99,"_100":101,"_127":128,"_135":-5,"_136":137,"_138":139,"_140":141,"_251":252,"_257":-5,"_108":235,"_258":137,"_111":259,"_260":261},"addedAt","2025-01-24T21:01:21.508439Z",[89,90,91,92,93,94,95,96],"category:browser-automation","category:agent-orchestration","environment:local","hosting:local-only","language:python","security_grade:a","quality_grade:a","license_grade:f","descriptionMarkdown","descriptionPlainText",737,"integrations",[102,113,120],{"_103":104,"_31":110,"_111":112},"brand",{"_105":106,"_37":107,"_108":109},"logoSvgPath","M12.48 10.92v3.28h7.84c-.24 1.84-.853 3.187-1.787 4.133-1.147 1.147-2.933 2.4-6.053 2.4-4.827 0-8.6-3.893-8.6-8.72s3.773-8.72 8.6-8.72c2.6 0 4.507 1.027 5.907 2.347l2.307-2.307C18.747 1.44 16.133 0 12.48 0 5.867 0 .307 5.387.307 12s5.56 12 12.173 12c3.573 0 6.267-1.173 8.373-3.36 2.16-2.16 2.84-5.213 2.84-7.667 0-.76-.053-1.467-.173-2.053H12.48z","Google","slug","google","Includes support for Google API integration as evidenced by the GOOGLE_API_KEY environment variable in the configuration.","uid","qv7epobont",{"_103":114,"_31":118,"_111":119},{"_105":115,"_37":116,"_108":117},"M16.361 10.26a.894.894 0 0 0-.558.47l-.072.148.001.207c0 .193.004.217.059.353.076.193.152.312.291.448.24.238.51.3.872.205a.86.86 0 0 0 .517-.436.752.752 0 0 0 .08-.498c-.064-.453-.33-.782-.724-.897a1.06 1.06 0 0 0-.466 0zm-9.203.005c-.305.096-.533.32-.65.639a1.187 1.187 0 0 0-.06.52c.057.309.31.59.598.667.362.095.632.033.872-.205.14-.136.215-.255.291-.448.055-.136.059-.16.059-.353l.001-.207-.072-.148a.894.894 0 0 0-.565-.472 1.02 1.02 0 0 0-.474.007Zm4.184 2c-.131.071-.223.25-.195.383.031.143.157.288.353.407.105.063.112.072.117.136.004.038-.01.146-.029.243-.02.094-.036.194-.036.222.002.074.07.195.143.253.064.052.076.054.255.059.164.005.198.001.264-.03.169-.082.212-.234.15-.525-.052-.243-.042-.28.087-.355.137-.08.281-.219.324-.314a.365.365 0 0 0-.175-.48.394.394 0 0 0-.181-.033c-.126 0-.207.03-.355.124l-.085.053-.053-.032c-.219-.13-.259-.145-.391-.143a.396.396 0 0 0-.193.032zm.39-2.195c-.373.036-.475.05-.654.086-.291.06-.68.195-.951.328-.94.46-1.589 1.226-1.787 2.114-.04.176-.045.234-.045.53 0 .294.005.357.043.524.264 1.16 1.332 2.017 2.714 2.173.3.033 1.596.033 1.896 0 1.11-.125 2.064-.727 2.493-1.571.114-.226.169-.372.22-.602.039-.167.044-.23.044-.523 0-.297-.005-.355-.045-.531-.288-1.29-1.539-2.304-3.072-2.497a6.873 6.873 0 0 0-.855-.031zm.645.937a3.283 3.283 0 0 1 1.44.514c.223.148.537.458.671.662.166.251.26.508.303.82.02.143.01.251-.043.482-.08.345-.332.705-.672.957a3.115 3.115 0 0 1-.689.348c-.382.122-.632.144-1.525.138-.582-.006-.686-.01-.853-.042-.57-.107-1.022-.334-1.35-.68-.264-.28-.385-.535-.45-.946-.03-.192.025-.509.137-.776.136-.326.488-.73.836-.963.403-.269.934-.46 1.422-.512.187-.02.586-.02.773-.002zm-5.503-11a1.653 1.653 0 0 0-.683.298C5.617.74 5.173 1.666 4.985 2.819c-.07.436-.119 1.04-.119 1.503 0 .544.064 1.24.155 1.721.02.107.031.202.023.208a8.12 8.12 0 0 1-.187.152 5.324 5.324 0 0 0-.949 1.02 5.49 5.49 0 0 0-.94 2.339 6.625 6.625 0 0 0-.023 1.357c.091.78.325 1.438.727 2.04l.13.195-.037.064c-.269.452-.498 1.105-.605 1.732-.084.496-.095.629-.095 1.294 0 .67.009.803.088 1.266.095.555.288 1.143.503 1.534.071.128.243.393.264.407.007.003-.014.067-.046.141a7.405 7.405 0 0 0-.548 1.873c-.062.417-.071.552-.071.991 0 .56.031.832.148 1.279L3.42 24h1.478l-.05-.091c-.297-.552-.325-1.575-.068-2.597.117-.472.25-.819.498-1.296l.148-.29v-.177c0-.165-.003-.184-.057-.293a.915.915 0 0 0-.194-.25 1.74 1.74 0 0 1-.385-.543c-.424-.92-.506-2.286-.208-3.451.124-.486.329-.918.544-1.154a.787.787 0 0 0 .223-.531c0-.195-.07-.355-.224-.522a3.136 3.136 0 0 1-.817-1.729c-.14-.96.114-2.005.69-2.834.563-.814 1.353-1.336 2.237-1.475.199-.033.57-.028.776.01.226.04.367.028.512-.041.179-.085.268-.19.374-.431.093-.215.165-.333.36-.576.234-.29.46-.489.822-.729.413-.27.884-.467 1.352-.561.17-.035.25-.04.569-.04.319 0 .398.005.569.04a4.07 4.07 0 0 1 1.914.997c.117.109.398.457.488.602.034.057.095.177.132.267.105.241.195.346.374.43.14.068.286.082.503.045.343-.058.607-.053.943.016 1.144.23 2.14 1.173 2.581 2.437.385 1.108.276 2.267-.296 3.153-.097.15-.193.27-.333.419-.301.322-.301.722-.001 1.053.493.539.801 1.866.708 3.036-.062.772-.26 1.463-.533 1.854a2.096 2.096 0 0 1-.224.258.916.916 0 0 0-.194.25c-.054.109-.057.128-.057.293v.178l.148.29c.248.476.38.823.498 1.295.253 1.008.231 2.01-.059 2.581a.845.845 0 0 0-.044.098c0 .006.329.009.732.009h.73l.02-.074.036-.134c.019-.076.057-.3.088-.516.029-.217.029-1.016 0-1.258-.11-.875-.295-1.57-.597-2.226-.032-.074-.053-.138-.046-.141.008-.005.057-.074.108-.152.376-.569.607-1.284.724-2.228.031-.26.031-1.378 0-1.628-.083-.645-.182-1.082-.348-1.525a6.083 6.083 0 0 0-.329-.7l-.038-.064.131-.194c.402-.604.636-1.262.727-2.04a6.625 6.625 0 0 0-.024-1.358 5.512 5.512 0 0 0-.939-2.339 5.325 5.325 0 0 0-.95-1.02 8.097 8.097 0 0 1-.186-.152.692.692 0 0 1 .023-.208c.208-1.087.201-2.443-.017-3.503-.19-.924-.535-1.658-.98-2.082-.354-.338-.716-.482-1.15-.455-.996.059-1.8 1.205-2.116 3.01a6.805 6.805 0 0 0-.097.726c0 .036-.007.066-.015.066a.96.96 0 0 1-.149-.078A4.857 4.857 0 0 0 12 3.03c-.832 0-1.687.243-2.456.698a.958.958 0 0 1-.148.078c-.008 0-.015-.03-.015-.066a6.71 6.71 0 0 0-.097-.725C8.997 1.392 8.337.319 7.46.048a2.096 2.096 0 0 0-.585-.041Zm.293 1.402c.248.197.523.759.682 1.388.03.113.06.244.069.292.007.047.026.152.041.233.067.365.098.76.102 1.24l.002.475-.12.175-.118.178h-.278c-.324 0-.646.041-.954.124l-.238.06c-.033.007-.038-.003-.057-.144a8.438 8.438 0 0 1 .016-2.323c.124-.788.413-1.501.696-1.711.067-.05.079-.049.157.013zm9.825-.012c.17.126.358.46.498.888.28.854.36 2.028.212 3.145-.019.14-.024.151-.057.144l-.238-.06a3.693 3.693 0 0 0-.954-.124h-.278l-.119-.178-.119-.175.002-.474c.004-.669.066-1.19.214-1.772.157-.623.434-1.185.68-1.382.078-.062.09-.063.159-.012z","Ollama","ollama","Supports integration with Ollama for local execution of Large Language Models, providing an alternative to cloud-based AI providers.","hg04ajg7fv",{"_103":121,"_31":125,"_111":126},{"_105":122,"_37":123,"_108":124},"M22.2819 9.8211a5.9847 5.9847 0 0 0-.5157-4.9108 6.0462 6.0462 0 0 0-6.5098-2.9A6.0651 6.0651 0 0 0 4.9807 4.1818a5.9847 5.9847 0 0 0-3.9977 2.9 6.0462 6.0462 0 0 0 .7427 7.0966 5.98 5.98 0 0 0 .511 4.9107 6.051 6.051 0 0 0 6.5146 2.9001A5.9847 5.9847 0 0 0 13.2599 24a6.0557 6.0557 0 0 0 5.7718-4.2058 5.9894 5.9894 0 0 0 3.9977-2.9001 6.0557 6.0557 0 0 0-.7475-7.0729zm-9.022 12.6081a4.4755 4.4755 0 0 1-2.8764-1.0408l.1419-.0804 4.7783-2.7582a.7948.7948 0 0 0 .3927-.6813v-6.7369l2.02 1.1686a.071.071 0 0 1 .038.052v5.5826a4.504 4.504 0 0 1-4.4945 4.4944zm-9.6607-4.1254a4.4708 4.4708 0 0 1-.5346-3.0137l.142.0852 4.783 2.7582a.7712.7712 0 0 0 .7806 0l5.8428-3.3685v2.3324a.0804.0804 0 0 1-.0332.0615L9.74 19.9502a4.4992 4.4992 0 0 1-6.1408-1.6464zM2.3408 7.8956a4.485 4.485 0 0 1 2.3655-1.9728V11.6a.7664.7664 0 0 0 .3879.6765l5.8144 3.3543-2.0201 1.1685a.0757.0757 0 0 1-.071 0l-4.8303-2.7865A4.504 4.504 0 0 1 2.3408 7.872zm16.5963 3.8558L13.1038 8.364 15.1192 7.2a.0757.0757 0 0 1 .071 0l4.8303 2.7913a4.4944 4.4944 0 0 1-.6765 8.1042v-5.6772a.79.79 0 0 0-.407-.667zm2.0107-3.0231l-.142-.0852-4.7735-2.7818a.7759.7759 0 0 0-.7854 0L9.409 9.2297V6.8974a.0662.0662 0 0 1 .0284-.0615l4.8303-2.7866a4.4992 4.4992 0 0 1 6.6802 4.66zM8.3065 12.863l-2.02-1.1638a.0804.0804 0 0 1-.038-.0567V6.0742a4.4992 4.4992 0 0 1 7.3757-3.4537l-.142.0805L8.704 5.459a.7948.7948 0 0 0-.3927.6813zm1.0976-2.3654l2.602-1.4998 2.6069 1.4998v2.9994l-2.5974 1.4997-2.6067-1.4997Z","OpenAI","openai","Integrates with OpenAI's models for language and vision capabilities, allowing the browser automation system to leverage OpenAI's AI models for processing and generating content.","o8n5ykadly","namespace",{"_129":130,"_108":134},"owner",{"_131":132},"githubAccount",{"_133":134},"login","JovaniPink","readmeMarkdown","recentUsage",0,"redditPostExternalId","1i95wtl","repository",{"_142":143,"_219":220,"_224":225,"_242":-5,"_243":244,"_245":-5,"_246":247},"argumentsJsonSchema",{"_144":145,"_216":217,"_149":218},"properties",{"_146":147,"_151":152,"_156":157,"_159":160,"_163":164,"_167":168,"_170":171,"_174":175,"_178":179,"_181":182,"_184":185,"_188":189,"_192":193,"_195":196,"_198":199,"_202":203,"_206":207,"_210":211,"_213":214},"CHROME_PATH",{"_31":148,"_149":150},"Optional: Path to Chrome executable","type","string","MCP_MAX_STEPS",{"_153":154,"_31":155,"_149":150},"default","30","Maximum number of steps","GOOGLE_API_KEY",{"_31":158,"_149":150},"Google API key","MCP_MODEL_NAME",{"_153":161,"_31":162,"_149":150},"claude-3-5-sonnet-20241022","Model name","MCP_USE_VISION",{"_153":165,"_31":166,"_149":150},"true","Enable vision capabilities","OPENAI_API_KEY",{"_31":169,"_149":150},"OpenAI API key","MCP_TEMPERATURE",{"_153":172,"_31":173,"_149":150},"0.3","Model temperature setting","OPENAI_ENDPOINT",{"_153":176,"_31":177,"_149":150},"https://api.openai.com/v1","OpenAI API endpoint URL","CHROME_USER_DATA",{"_31":180,"_149":150},"Optional: Chrome user data directory","ANTHROPIC_API_KEY",{"_31":183,"_149":150},"Anthropic API key","MCP_MODEL_PROVIDER",{"_153":186,"_31":187,"_149":150},"anthropic","Model provider (options: anthropic, openai, azure, deepseek)","ANONYMIZED_TELEMETRY",{"_153":190,"_31":191,"_149":150},"false","Set to false to disable anonymized telemetry","AZURE_OPENAI_API_KEY",{"_31":194,"_149":150},"Azure OpenAI API key","AZURE_OPENAI_ENDPOINT",{"_31":197,"_149":150},"Azure OpenAI endpoint URL","CHROME_DEBUGGING_HOST",{"_153":200,"_31":201,"_149":150},"localhost","Chrome debugging host","CHROME_DEBUGGING_PORT",{"_153":204,"_31":205,"_149":150},"9222","Chrome debugging port","MCP_MAX_ACTIONS_PER_STEP",{"_153":208,"_31":209,"_149":150},"5","Maximum actions per step","MCP_TOOL_CALL_IN_CONTENT",{"_153":165,"_31":212,"_149":150},"Enable tool calls in content","CHROME_PERSISTENT_SESSION",{"_153":190,"_31":215,"_149":150},"Set to true to keep browser open between AI tasks","required",[],"object","githubProject",{"_35":221,"_222":223},"# MCP Browser Use Server\n\n[![smithery badge](https://smithery.ai/badge/@JovaniPink/mcp-browser-use)](https://smithery.ai/server/@JovaniPink/mcp-browser-use)\n\n> Model Context Protocol (MCP) server that wires [browser-use](https://github.com/browser-use/browser-use) into Claude Desktop and other MCP compatible clients.\n\n $\"Browser$ \n\n## Overview\n\nThis repository provides a production-ready wrapper around the `browser-use` automation engine. It exposes a single MCP tool (`run_browser_agent`) that orchestrates a browser session, executes the `browser-use` agent, and returns the final result back to the client. The refactored layout focuses on keeping configuration in one place, improving testability, and keeping `browser-use` upgrades isolated from MCP specific code.\n\n### Key Capabilities\n\n- **Automated browsing** – Navigate, interact with forms, control tabs, capture screenshots, and read page content through natural-language instructions executed by `browser-use`.\n- **Agent lifecycle management** – `CustomAgent` wraps `browser-use`'s base agent to add history export, richer prompts, and consistent error handling across runs.\n- **Centralised browser configuration** – `create_browser_session` translates environment variables into a ready-to-use `BrowserSession`, enabling persistent profiles, proxies, and custom Chromium flags without touching the agent logic.\n- **FastMCP integration** – `server.py` registers the MCP tool, normalises configuration, and ensures the browser session is always cleaned up.\n- **Client helpers** – `client.py` includes async helpers for tests or other Python processes that wish to exercise the MCP server in-process.\n\n### Project Structure\n\n```\n.\n├── documentation/\n│ ├── CONFIGURATION.md # Detailed configuration reference\n│ └── SECURITY.md # Security considerations for running the server\n├── .env.example # Example environment variables for local development\n├── src/mcp_browser_use/\n│ ├── agent/ # Custom agent, prompts, message history, and views\n│ ├── browser/ # Browser session factory and persistence helpers\n│ ├── controller/ # Custom controller extensions for clipboard actions\n│ ├── utils/ # LLM factory, agent state helpers, encoding utilities\n│ ├── client.py # Async helper for connecting to the FastMCP app\n│ └── server.py # FastMCP app and the `run_browser_agent` tool\n└── tests/ # Unit tests covering server helpers and agent features\n```\n\n## Getting Started\n\n### Requirements\n\n- Python 3.11+\n- Google Chrome or Chromium (for local automation)\n- [`uv`](https://github.com/astral-sh/uv) for dependency management (recommended)\n- Optional: Claude Desktop or another MCP-compatible client for integration testing\n\n### Installation\n\n```bash\ngit clone https://github.com/JovaniPink/mcp-browser-use.git\ncd mcp-browser-use\nuv sync\n```\n\nCopy `sample.env` to `.env` (or export the variables in another way) and update the values for the providers you plan to use.\n\n### Launching the server\n\n```bash\nuv run mcp-browser-use\n```\n\nThe command invokes the console script defined in `pyproject.toml`, starts the FastMCP application, and registers the `run_browser_agent` tool.\n\n#### Using with Claude Desktop\n\nOnce the server is running you can register it inside Claude Desktop, for example:\n\n```json\n\"mcpServers\": {\n \"mcp_server_browser_use\": {\n \"command\": \"uvx\",\n \"args\": [\"mcp-browser-use\"],\n \"env\": {\n \"MCP_MODEL_PROVIDER\": \"anthropic\",\n \"MCP_MODEL_NAME\": \"claude-3-5-sonnet-20241022\"\n }\n }\n}\n```\n\n### Debugging\n\nFor interactive debugging, use the [MCP Inspector](https://github.com/modelcontextprotocol/inspector):\n\n```bash\nnpx @modelcontextprotocol/inspector uv --directory /path/to/project run mcp-browser-use\n```\n\nThe inspector prints a URL that can be opened in the browser to watch tool calls and responses in real time.\n\n## Configuration\n\nA full list of environment variables and their defaults is available in [documentation/CONFIGURATION.md](documentation/CONFIGURATION.md). Highlights include:\n\n- `MCP_MODEL_PROVIDER`, `MCP_MODEL_NAME`, `MCP_TEMPERATURE`, `MCP_MAX_STEPS`, `MCP_MAX_ACTIONS_PER_STEP`, and `MCP_USE_VISION` control the LLM and agent run.\n- Provider-specific API keys and endpoints (`ANTHROPIC_API_KEY`, `OPENAI_API_KEY`, `DEEPSEEK_API_KEY`, `GOOGLE_API_KEY`, `AZURE_OPENAI_API_KEY`, etc.).\n- Browser runtime flags (`BROWSER_USE_HEADLESS`, `BROWSER_USE_EXTRA_CHROMIUM_ARGS`, `CHROME_PERSISTENT_SESSION`, `BROWSER_USE_PROXY_URL`, ...).\n\nUse `.env` + [`python-dotenv`](https://pypi.org/project/python-dotenv/) or your preferred secrets manager to keep credentials out of source control.\n\n## Running Tests\n\n```bash\nuv run pytest\n```\n\nThe tests cover the custom agent behaviour, browser session factory, and other utility helpers.\n\n## Security\n\nControlling a full browser instance remotely can grant broad access to the host machine. Review [documentation/SECURITY.md](documentation/SECURITY.md) before exposing the server to untrusted environments.\n\n## Contributing\n\n1. Fork the repository\n2. Create your feature branch: `git checkout -b my-new-feature`\n3. Commit your changes: `git commit -m 'Add some feature'`\n4. Push to the branch: `git push origin my-new-feature`\n5. Open a pull request\n\nBug reports and feature suggestions are welcome—please include logs and reproduction steps when applicable.\n","url","https://github.com/JovaniPink/mcp-browser-use","githubRepository",{"_226":227,"_228":229,"_230":231,"_16":232,"_233":234,"_37":235,"_129":236,"_238":137,"_239":-5,"_240":241},"defaultBranch","main","fullName","JovaniPink/mcp-browser-use","githubTreeId",439667,1078,"language","Python","mcp-browser-use",{"_133":134,"_111":237},"05fxu6vay2","recentStargazers","spdxLicense","stargazers",55,"glamaJson","inspectable",true,"npmPackage","supportedPlatforms",[248,249,250],"LINUX","MACOS","WINDOWS","scores",{"_253":137,"_254":255,"_256":255},"license","quality",100,"security","simpleIcon","toolCount","tjea5rgnbv","updatedAt","2025-11-17T15:35:52.827686Z","meta",[264,267,269,276,279,281,284,287,290,293,296,298,301,304,306,308,310,312,314],{"_265":266},"title","MCP Browser Use Server | Glama",{"_268":43,"_37":31},"content",{"_270":271,"_272":273,"_274":275},"href","https://glama.ai/mcp/servers/@JovaniPink/mcp-browser-use","rel","canonical","tagName","link",{"_268":44,"_277":278},"property","og:title",{"_268":43,"_277":280},"og:description",{"_268":282,"_277":283},"https://glama.ai/logo.png","og:logo",{"_268":285,"_277":286},"https://glama.ai/mcp/servers/@JovaniPink/mcp-browser-use/og-image","og:image",{"_268":288,"_277":289},"630","og:image:height",{"_268":291,"_277":292},"1200","og:image:width",{"_268":294,"_277":295},"website","og:type",{"_268":271,"_277":297},"og:url",{"_268":299,"_277":300},"Glama – MCP Hosting Platform","og:site_name",{"_268":302,"_37":303},"summary_large_image","twitter:card",{"_268":299,"_37":305},"twitter:title",{"_268":43,"_37":307},"twitter:description",{"_268":285,"_37":309},"twitter:image",{"_268":288,"_37":311},"twitter:image:height",{"_268":291,"_37":313},"twitter:image:width",{"_315":316},"script:ld+json",{"_317":318,"_319":320},"@context","http://schema.org","@graph",[321],{"_317":322,"_323":324,"_325":326},"https://schema.org","@type","BreadcrumbList","itemListElement",[327,334,337,341],{"_323":328,"_329":330,"_37":331,"_332":333},"ListItem","item","https://glama.ai/","Glama","position",1,{"_323":328,"_329":335,"_37":336,"_332":22},"https://glama.ai/mcp","MCP",{"_323":328,"_329":338,"_37":339,"_332":340},"https://glama.ai/mcp/servers","Servers",3,{"_323":328,"_329":271,"_37":44,"_332":342},4,"visitorIsMaintainer",false,"routes/_public/mcp/servers/~namespace/~slug/_pages/_explorer/blob/~commit/~/_route",{"_347":348,"_29":354,"_394":395,"_262":408,"_406":432,"_433":434},"blob",{"_268":349,"_350":351,"_352":353},"class SendKeysEvent:\n def __init__(self, keys: str):\n self.keys = keys\n\n\nclass ScreenshotEvent:\n def __init__(self, full_page: bool = False):\n self.full_page = full_page\n","fileName","events.py","size",190,{"_86":87,"_13":355,"_97":43,"_98":43,"_33":44,"_16":99,"_100":356,"_127":363,"_135":-5,"_136":137,"_138":139,"_140":366,"_251":393,"_257":-5,"_108":235,"_258":137,"_111":259,"_260":261},[89,90,91,92,93,94,95,96],[357,359,361],{"_103":358,"_31":110,"_111":112},{"_105":106,"_37":107,"_108":109},{"_103":360,"_31":118,"_111":119},{"_105":115,"_37":116,"_108":117},{"_103":362,"_31":125,"_111":126},{"_105":122,"_37":123,"_108":124},{"_129":364,"_108":134},{"_131":365},{"_133":134},{"_142":367,"_219":389,"_224":390,"_242":-5,"_243":244,"_245":-5,"_246":392},{"_144":368,"_216":388,"_149":218},{"_146":369,"_151":370,"_156":371,"_159":372,"_163":373,"_167":374,"_170":375,"_174":376,"_178":377,"_181":378,"_184":379,"_188":380,"_192":381,"_195":382,"_198":383,"_202":384,"_206":385,"_210":386,"_213":387},{"_31":148,"_149":150},{"_153":154,"_31":155,"_149":150},{"_31":158,"_149":150},{"_153":161,"_31":162,"_149":150},{"_153":165,"_31":166,"_149":150},{"_31":169,"_149":150},{"_153":172,"_31":173,"_149":150},{"_153":176,"_31":177,"_149":150},{"_31":180,"_149":150},{"_31":183,"_149":150},{"_153":186,"_31":187,"_149":150},{"_153":190,"_31":191,"_149":150},{"_31":194,"_149":150},{"_31":197,"_149":150},{"_153":200,"_31":201,"_149":150},{"_153":204,"_31":205,"_149":150},{"_153":208,"_31":209,"_149":150},{"_153":165,"_31":212,"_149":150},{"_153":190,"_31":215,"_149":150},[],{"_35":221,"_222":223},{"_226":227,"_228":229,"_230":231,"_16":232,"_233":234,"_37":235,"_129":391,"_238":137,"_239":-5,"_240":241},{"_133":134,"_111":237},[248,249,250],{"_253":137,"_254":255,"_256":255},"member",{"_396":397,"_400":401,"_402":403,"_230":404,"_405":344,"_406":351,"_407":244,"_352":353,"_149":347},"commit",{"_398":399},"sha","e5bc348ab030fac9a8965731e16aea0117b7c523","contentStoredObjectPointerId",13343119,"fullPath","tests/stubs/browser_use/browser/events.py",439654,"indexable","path","safeToDisplayAsText",[409,411,413,415,417,418,419,421,422,423,424,425,426,427,428,429,430,431],{"_265":410},"events.py | MCP Browser Use Server | Glama",{"_268":412,"_37":31},"Enterprise-grade security, privacy, and features like agents, MCP, prompt templates, and more.",{"_270":414,"_272":273,"_274":275},"https://glama.ai/mcp/servers/@JovaniPink/mcp-browser-use/blob/e5bc348ab030fac9a8965731e16aea0117b7c523/tests/stubs/browser_use/browser",{"_268":416,"_277":278},"events.py | MCP Browser Use Server",{"_268":412,"_277":280},{"_268":282,"_277":283},{"_268":420,"_277":286},"https://glama.ai/generated-images/og?title=events.py+%7C+MCP+Browser+Use+Server",{"_268":288,"_277":289},{"_268":291,"_277":292},{"_268":294,"_277":295},{"_268":414,"_277":297},{"_268":299,"_277":300},{"_268":302,"_37":303},{"_268":299,"_37":305},{"_268":412,"_37":307},{"_268":420,"_37":309},{"_268":288,"_37":311},{"_268":291,"_37":313},"tests/stubs/browser_use/browser","treeMembers",[435,439,443,445,449,453],{"_396":436,"_402":437,"_406":438,"_149":347},{"_398":399},"tests/stubs/browser_use/browser/browser.py","browser.py",{"_396":440,"_402":441,"_406":442,"_149":347},{"_398":399},"tests/stubs/browser_use/browser/context.py","context.py",{"_396":444,"_402":403,"_406":351,"_149":347},{"_398":399},{"_396":446,"_402":447,"_406":448,"_149":347},{"_398":399},"tests/stubs/browser_use/browser/__init__.py","__init__.py",{"_396":450,"_402":451,"_406":452,"_149":347},{"_398":399},"tests/stubs/browser_use/browser/profile.py","profile.py",{"_396":454,"_402":455,"_406":456,"_149":347},{"_398":399},"tests/stubs/browser_use/browser/views.py","views.py","actionData","errors"]

MCP directory API